Download numerical analysis, design and two port equivalent circuit

Document related concepts

Fault tolerance wikipedia , lookup

Electrical substation wikipedia , lookup

Mathematics of radio engineering wikipedia , lookup

Non-radiative dielectric waveguide wikipedia , lookup

Chirp spectrum wikipedia , lookup

Regenerative circuit wikipedia , lookup

History of electric power transmission wikipedia , lookup

Distributed element filter wikipedia , lookup

Resonant inductive coupling wikipedia , lookup

Transmission line loudspeaker wikipedia , lookup

Array processing wikipedia , lookup

RLC circuit wikipedia , lookup

Two-port network wikipedia , lookup

Transcript
NUMERICAL ANALYSIS, DESIGN AND TWO PORT EQUIVALENT
CIRCUIT MODELS FOR SPLIT RING RESONATOR ARRAYS
A THESIS SUBMITTED TO
THE GRADUATE SCHOOL OF NATURAL AND APPLIED SCIENCES
OF
MIDDLE EAST TECHNICAL UNIVERSITY
BY
PINAR YAŞAR ÖRTEN
IN PARTIAL FULFILLMENT OF THE REQUIREMENTS
FOR
THE DEGREE OF MASTER OF SCIENCE
IN
ELECTRICAL AND ELECTRONICS ENGINEERING
FEBRUARY 2010
Approval of the thesis:
NUMERICAL ANALYSIS, DESIGN AND TWO PORT EQUIVALENT
CIRCUIT MODELS FOR SPLIT RING RESONATOR ARRAYS
submitted by PINAR YAŞAR ÖRTEN in partial fulfillment of the requirements
for the degree of Master of Science in Electrical and Electronics Engineering
Department, Middle East Technical University by,
Prof. Dr. Canan Özgen
Dean, Graduate School of Natural and Applied Sciences
Prof. Dr. İsmet Erkmen
Head of Department, Electrical and Electronics Engineering
Prof. Dr. Gönül Turhan Sayan
Supervisor, Electrical and Electronics Engineering Dept., METU
Examining Committee Members:
Prof. Dr. Gülbin Dural
Electrical and Electronics Engineering Dept., METU
Prof. Dr. Gönül Turhan Sayan
Electrical and Electronics Engineering Dept., METU
Prof. Dr. Mustafa Kuzuoğlu
Electrical and Electronics Engineering Dept., METU
Prof. Dr. Kemal Leblebicioğlu
Electrical and Electronics Engineering Dept., METU
Prof. Dr. Adnan Köksal
Electrical and Electronics Engineering Dept., Hacettepe Univ.
Date:
04.02.2010
I hereby declare that all information in this document has been obtained and
presented in accordance with academic rules and ethical conduct. I also
declare that, as required by these rules and conduct, I have fully cited and
referenced all material and results that are not original to this work.
Name, Last name
: Pınar YAŞAR ÖRTEN
Signature
:
iii
ABSTRACT
NUMERICAL ANALYSIS, DESIGN AND TWO PORT EQUIVALENT
CIRCUIT MODELS FOR SPLIT RING RESONATOR ARRAYS
Yaşar Örten, Pınar
M. Sc., Department of Electrical and Electronics Engineering
Supervisor: Prof. Dr. Gönül Turhan Sayan
February 2010, 95 pages
Split ring resonator (SRR) is a metamaterial structure which displays negative
permeability values over a relatively small bandwidth around its magnetic
resonance frequency. Unit SRR cells and arrays have been used in various novel
applications including the design of miniaturized microwave devices and antennas.
When the SRR arrays are combined with the arrays of conducting wires, left handed
materials can be constructed with the unusual property of having negative valued
effective refractive indices.
In this thesis, unit cells and arrays of single-ring multiple-split type SRR structures
are numerically analyzed by using Ansoft’s HFSS software that is based on the
finite elements method (FEM). Some of these structures are constructed over lowloss dielectric substrates and their complex scattering parameters are measured to
verify the numerical simulation results. The major purpose of this study has been to
establish equivalent circuit models to estimate the behavior of SRR structures in a
simple and computationally efficient manner. For this purpose, individual single
ring SRR cells with multiple splits are modeled by appropriate two-port RLC
resonant circuits paying special attention to conductor and dielectric loss effects.
Results obtained from these models are compared with the results of HFSS
iv
simulations which use either PEC/PMC (perfect electric conductor/perfect magnetic
conductor) type or perfectly matched layer (PML) type boundary conditions.
Interactions between the elements of SRR arrays such as the mutual inductance and
capacitance effects as well as additional dielectric losses are also modeled by proper
two-port equivalent circuits to describe the overall array behavior and to compute
the associated transmission spectrum by simple MATLAB codes. Results of
numerical HFSS simulations, equivalent circuit model computations and
measurements are shown to be in good agreement.
Keywords: Split ring resonator (SRR), metamaterials, two-port networks,
equivalent circuit modeling, resonance frequency, HFSS simulations.
v
ÖZ
YARIKLI HALKA REZONATÖR DİZİLERİNİN SAYISAL ANALİZİ,
TASARIMI VE İKİ KAPILI EŞDEĞER DEVRE MODELLEMESİ
Yaşar Örten, Pınar
Yüksek Lisans, Elektrik ve Elektronik Mühendisliği Bölümü
Tez Yöneticisi: Prof. Dr. Gönül Turhan Sayan
Şubat 2010, 95 sayfa
Yarıklı halka rezonatörü (YHR) manyetik rezonans frekansı etrafındaki göreceli
olarak dar bir bantta etkin negative manyetik geçirgenlik özelliği gösteren bir
metamalzeme yapısıdır. YHR birim hücre ve dizileri, küçük boyutlu mikrodalga
gereçleri ve antenlerin tasarımını da içeren çeşitli özgün uygulamalarda
kullanılmaktadırlar. YHR dizilerinin iletken tel dizileri ile birlikte kullanılmasıyla,
olağan dışı bir şekilde, negatif etkin kırılma indisine sahip olma özelliği gösteren
sol-elli malzemeler tasarlanabilmektedir.
Bu tezde, çoklu yarıklı tek halkalı YHR birim hücreleri ve dizileri, sonlu elemanlar
yöntemine dayalı bir yöntem kullanan Ansoft firmasının HFSS yazılımı ile sayısal
olarak analiz edilmiştir. Bu yapıların bazıları düşük kayıplı dielektrik levhalar
üzerinde gerçekleştirilmiş ve sayısal benzetim sonuçlarını doğrulamak için
kompleks S-parametreleri ölçülmüştür. Bu çalışmanın esas amacı, YHR yapılarının
davranışını basit ve etkin bir şekilde tahmin edip hesaplayabilmek için eşdeğer
devre modelleri kurmaktır. Bu nedenle, çoklu yarıklı tek halkalı YHR birim
hücreleri ve dizileri, iletken ve dielektrik bölgelerden kaynaklanan kayıp etkilerinin
de özenle hesaba katıldığı uygun iki kapılı RLC rezonans devreleri ile
modellenmiştir. Bu modellerden elde edilen sonuçlar, sınır koşul olarak mükemmel
vi
elektrik iletken/mükemmel manyetik iletken veya mükemmel uyumlu tabaka tipi
sınır koşullarının kullanıldığı HFSS benzetim sonuçları ile karşılaştırılmıştır. YHR
dizilerinin elemanları arasındaki karşılıklı endüktans, kapasitans etkileri ve ilave
dielektrik kayıplar gibi etkileşimler uygun iki kapılı eşdeğer devre yapıları ile
modellenerek dizi davranışları tahmin edilmiş ve ilgili dizi yapısının iletim
spektrumu
yalın
MATLAB
kodları
ile
hesaplanmıştır.
Sayısal
HFSS
benzetimlerinden, eşdeğer devre modellerinin hesaplamalarından ve ölçümlerden
elde edilen sonuçların birbirleriyle uyumlu oldukları gösterilmiştir.
Anahtar kelimeler: Yarıklı halka rezonatörü (YHR), metamalzemeler, iki kapılı
devreler, eşdeğer devre modellemesi, rezonans frekansı, HFSS simülasyonları.
vii
Canım Aileme
viii
ACKNOWLEDGEMENTS
I would like to express my gratitude to my supervisor Prof. Dr. Gönül Turhan
Sayan not only for her invaluable guidance but also for helping me to be optimistic
even at the most difficult times of this thesis research. I have learned lots of things
from her both about engineering and life.
I would like to thank the members of my thesis committee, Prof Dr. Gülbin Dural,
Prof Dr. Mustafa Kuzuoğlu, Prof. Dr. Kemal Leblebicioğlu and Prof. Dr. Adnan
Köksal for reading the manuscript and sharing their opinions. I also would like to
thank Evren Ekmekçi for his helps especially during the experimental work.
I also would like to thank ASELSAN A.Ş. for the facilities provided during my
graduate degree and TÜBITAK-BİDEB (The Scientific and Technological
Research Council of Turkey-The Department of Science Fellowships and Grant
Programmes) for its support to scientific research.
Thanks to all of my close friends for their support and encouragement.
Lastly but mostly, I would like to thank my mother, father and sister for giving me a
great and special family. Without their support, I would not be able to overcome the
difficulties throughout my life. And special thanks to my spouse Hasan who has
never given up believing in me and even at the hardest times made me smile to life.
I love all of you so much my dear family.
ix
TABLE OF CONTENTS
ABSTRACT.............................................................................................................. iv
ÖZ…… ..................................................................................................................... vi
ACKNOWLEDGEMENTS ...................................................................................... ix
ACKNOWLEDGEMENTS ...................................................................................... ix
TABLE OF CONTENTS........................................................................................... x
LIST OF TABLES .................................................................................................. xiii
LIST OF FIGURES ................................................................................................ xiv
CHAPTER
1. INTRODUCTION ................................................................................................. 1
2. BASICS OF NUMERICAL SIMULATIONS AND EQUIVALENT CIRCUIT
MODELING FOR SRR STRUCTURES .................................................................. 6
2.1
Introduction ................................................................................................ 6
2.2
Use of HFSS Simulations to Compute the Transmission Spectra of SRR
Structures ............................................................................................................... 7
2.2.1 Simulation Problem 1 ............................................................................ 8
2.2.2 Simulation Problem 2 .......................................................................... 12
2.2.3 Simulation Problem 3 .......................................................................... 17
2.3
Basics of SRR Modeling by Equivalent Lumped Circuit Models ........... 20
2.3.1 Modeling an SRR Unit Cell by a Proper RLC Circuit ........................ 21
x
3. NUMERICAL SIMULATIONS AND TWO PORT EQUIVALENT CIRCUIT
MODELING OF SQUARE SHAPED SINGLE LOOP SRR STRUCTURES ....... 36
3.1
Introduction .............................................................................................. 36
3.2
Single-Loop Square-Shaped SRR Unit Cell ............................................ 37
3.2.1 Numerical Simulation and Equivalent Circuit Modeling of the Isolated
SRR Unit Cell .................................................................................................. 37
3.2.2 Numerical Simulation and Equivalent Circuit Modeling of the Infinite
SRR Array along the Electric Field Direction ................................................. 43
3.3
An SRR Array Extending Along the Propagation Direction ................... 48
3.3.1 Analysis of the Isolated Two-Element Array of Size 2x1x1 ............... 49
3.3.2 Numerical Simulation and Equivalent Circuit Modeling of Two Layer
Infinite SRR Array ........................................................................................... 54
3.4
An SRR Array Extending Along the Electric Field Direction................. 58
3.4.1 Analysis of the Isolated Two-Element Array of Size 1x2x1 ............... 59
3.4.2 Numerical Simulation and Equivalent Circuit Modeling of Infinite SRR
Array Along the Electric Field Direction ........................................................ 63
3.5
The 2x2x1 SRR Array ............................................................................. 64
3.5.1 Analysis of the Isolated Four Element Array of Size 2x2x1 ............... 67
3.5.2 Numerical Simulation and Equivalent Circuit Modeling of Infinite Two
Layer SRR Array ............................................................................................. 71
4. NUMERICAL SIMULATIONS, MEASUREMENTS AND TWO PORT
EQUIVALENT CIRCUIT MODELING OF SQUARE SHAPED SINGLE LOOP
SRR STRUCTURES ............................................................................................... 72
4.1
Introduction .............................................................................................. 72
4.2
Analysis and Measurements of the SRR Unit Cell .................................. 73
4.2.1 Analysis of the Isolated SRR Unit Cell ............................................... 74
4.2.2 Numerical Simulation, Equivalent Circuit Modeling and Measurement
of the SRR Unit Cell within Waveguide ......................................................... 76
xi
4.3
Analysis and Measurements of the 4x1x1 SRR Array ............................ 80
4.3.1 Analysis of the Isolated 4x1x1 SRR Array .......................................... 81
4.3.2 Measurements and Numerical Simulations for the 4x1x1 SRR Array
Placed Within a Waveguide ............................................................................. 85
5. CONCLUSIONS AND FUTURE WORK .......................................................... 89
REFERENCES ........................................................................................................ 93
xii
LIST OF TABLES
TABLES
Table 2-1: Comparison of mesh numbers used in HFSS simulations with PML
boundary conditions for different number of array elements in electric field
direction ................................................................................................................... 19
Table 3-1: Geometrical parameters of the SRR unit cell shown in Figure 3-2........ 39
Table 3-2: Calculated values of circuit parameters for the SRR unit cell ............... 40
Table 3-3: Calculated circuit model parameters of the infinite SRR array extending
in the electric field direction .................................................................................... 47
Table 3-4: Calculated model parameters for the infinite SRR array in electric field
direction ................................................................................................................... 58
Table 4-1: Geometrical parameters of the fabricated SRR unit cell ....................... 74
Table 4-2: Model parameters of the fabricated SRR unit cell ................................. 75
Table 4-3: Calculated model parameters of the infinite SRR array extending along
the incident electric field direction .......................................................................... 79
xiii
LIST OF FIGURES
FIGURES
Figure 1-1: Unit cell geometry for a) Conventional SRR, b) BC-SRR. .................... 2
Figure 2-1: Notation used to identify SRR array topologies. .................................... 6
Figure 2-2: Schematic views of SRR unit cell and the HFSS simulation set-up a)
Geometry and dimensions of the SRR unit cell, b) Boundary conditions used for
HFSS simulations. ..................................................................................................... 9
Figure 2-3: Two dimensional periodic SRR array implemented by the use of PEC
and PMC boundary conditions in HFSS simulation a) Array in E field direction, b)
Array in H field direction......................................................................................... 10
Figure 2-4: Effect of vacuum length in H field direction on resonance frequency
(red curve D H =1.75 mm, blue curve D H =3.5 mm). .............................................. 12
Figure 2-5: Variation of transmission minimum with substrate length in electric
field direction. .......................................................................................................... 13
Figure 2-6: Variation of mutual inductance with cell to cell separation distance
along the electric field direction. ............................................................................. 14
Figure 2-7: Variation of mutual capacitance with cell to cell separation distance
along the electric field direction. ............................................................................. 15
Figure 2-8: Rate of change of mutual inductance with d. ........................................ 16
Figure 2-9: Rate of change of mutual capacitance with d. ...................................... 16
Figure 2-10: Variation of resonance frequency with d based on calculated circuit
parameters. ............................................................................................................... 17
Figure 2-11: Unit cell geometry used for comparison of boundary conditions. ...... 18
Figure 2-12: Comparison of transmission spectrum of different SRR configurations
with PEC and PML boundaries using DE=1 mm and DH=11.049 mm. ................... 19
Figure 2-13: Two-port equivalent circuit representation of a single loop SRR cell
with either series or parallel LC resonant circuit in the shunt branch. .................... 22
xiv
Figure 2-14: Two different representations for the impedance Z a) Series LC circuit,
b) Parallel LC circuit................................................................................................ 25
Figure 2-15: A feasible two-port equivalent circuit representation for SRR unit cell
including ohmic loss effects. ................................................................................... 27
Figure 2-16: Transmission spectrum curves of the isolated SRR cell, which are
obtained by HFSS simulation with PML boundary conditions (with DE=1 mm and
DH=5 mm) and by equivalent circuit modeling approach. Geometry and dimensions
for the SRR cell are described in Section 2.2.3. Equivalent circuit model is given in
Figure 2-15. .............................................................................................................. 29
Figure 2-17: Alternative two-port equivalent circuit representation of a single loop
SRR with either series or parallel LC resonant circuit in the series branch. ........... 30
Figure 2-18: Alternative representations of two-port network consisting of series
impedance a) Series LC circuit, b) Parallel LC circuit. ......................................... 33
Figure 2-19: Improved version of the two-port circuit representation shown in
Figure 2-18 (b) with the inclusion of equivalent loss resistances. ........................... 34
Figure 2-20: Transmission spectrum curves of the isolated SRR cell, which are
obtained by HFSS simulation with PML boundary conditions (with DE=1 mm and
DH=5 mm) and by equivalent circuit modeling approach. Geometry and dimensions
for the SRR cell is described in Section 2.2.3. Equivalent circuit model is given in
Figure 2-19. .............................................................................................................. 35
Figure 3-1: Boundary conditions for HFSS simulations: a) PML boundary
conditions to examine an isolated SRR unit cell, b) PEC/PMC boundary conditions
to examine a two dimensional infinite SRR array. .................................................. 37
Figure 3-2: a) Unit cell geometry of a single square loop SRR, b) Parameters of the
isolated SRR cell, c) Equivalent two-port circuit model of the SRR cell. ............... 39
Figure 3-3: Estimated transmission spectra of the SRR unit cell using HFSS
simulations with PML boundary conditions and DH=2 mm and using the equivalent
two-port circuit given in Figure 3-2 (c). .................................................................. 42
Figure 3-4: Estimated transmission spectra of the SRR unit cell using HFSS
simulations with PML boundary conditions and DH=1 mm and using the equivalent
two-port circuit given in Figure 3-2 (c). .................................................................. 43
xv
Figure 3-5: Equivalent circuit model for the SRR array which effectively extends to
infinity in the electric field direction. ...................................................................... 45
Figure 3-6: Transmission spectrum of the infinite SRR array as estimated by an
HFSS simulation using the PEC/PMC boundary conditions. .................................. 48
Figure 3-7: a) Geometry of 2x1x1 SRR array, b) Elements of 2x1x1 array, c) Twoport representation of 2x1x1 array (First alternative for the coupling two-port
connection)............................................................................................................... 49
Figure 3-8: Estimated transmission spectrum of 2x1x1 SRR array using HFSS
simulations with PML boundary conditions and DH=1 mm and using the equivalent
two-port circuit given in Figure 3-7(c). ................................................................... 52
Figure 3-9: Two-port representation of the 2x1x1 array (Second alternative for the
coupling two-port connection). ................................................................................ 53
Figure 3-10: Estimated transmission spectrum of the 2x1x1 SRR array using HFSS
simulations with PML boundary conditions and DH=1 mm. and using the equivalent
two-port circuit given in Figure 3-9. ........................................................................ 54
Figure 3-11: An equivalent circuit model for the double layer SRR array (m x n x p)
where m=2, p=1 (due to sparse array approximation) and n approaches to infinity.
................................................................................................................................. 56
Figure 3-12: Simulation of 2x1x1 array with PEC/PMC boundary. ....................... 58
Figure 3-13: a) Geometry of the 1x2x1 array, b) Elements of 1x2x1 array, c)
Equivalent two-port circuit model for the 1x2x1 array. .......................................... 60
Figure 3-14: Estimated transmission spectra of the 1x2x1 SRR array using HFSS
simulations with PML boundary conditions and D H =1.75 mm and using the
equivalent two-port circuit given in Figure 3-13 (c)................................................ 63
Figure 3-15: Transmission spectrum of the 1x2x1 SRR array obtained by HFSS
simulation with PEC/PMC boundary conditions. .................................................... 64
Figure 3-16: a) Unit cell geometry of 2x2x1 array, b) Parameters of the 2x2x1 SRR
array. ........................................................................................................................ 66
Figure 3-17: Two-port model for 2x2x1 SRR array. ............................................... 68
xvi
Figure 3-18: Estimated transmission spectra of the 2x2x1 SRR array using HFSS
simulations with PML boundary conditions and D H =1.75 mm and using the
equivalent circuit model shown in Figure 3-17. ...................................................... 70
Figure 3-19: Transmission spectrum for the 2x2x1 SRR array estimated by HFSS
with PEC/PMC boundary conditions. ...................................................................... 71
Figure 4-1: Boundary conditions used in HFSS simulations: a) Use of PML
conditions to examine the isolated SRR cell, b) Use of PEC boundary conditions to
simulate SRR behavior within a metallic waveguide. ............................................. 73
Figure 4-2: SRR unit cell a) Geometry and excitation, b) Design parameters. ....... 74
Figure 4-3: Equivalent two-port circuit representation of the SRR unit cell. .......... 75
Figure 4-4: Estimated transmission spectra of the SRR unit cell using HFSS
simulations with PML boundary conditions and DH=5 mm. and using the equivalent
two-port circuit given in Figure 4-3. ........................................................................ 76
Figure 4-5: Equivalent circuit model of the effective SRR array formed by placing
SRR unit cell within the waveguide in measurements. ........................................... 78
Figure 4-6: Simulated transmission spectrum of the fabricated SRR unit cell using
HFSS with PEC boundary conditions. ..................................................................... 79
Figure 4-7: Measured transmission spectrum of the fabricated SRR unit cell when it
is placed within a measurement waveguide. ............................................................ 80
Figure 4-8: Four element SRR array of size 4x1x1 a) Geometry and excitation, b)
Capacitive coupling effects along the propagation direction................................... 81
Figure 4-9: Two-port equivalent circuit model suggested for the isolated 4x1x1
SRR array. ................................................................................................................ 83
Figure 4-11: Estimated transmission spectrum of the isolated 4x1x1 SRR array
using HFSS with PML boundary conditions and using the equivalent two-port
model given in Figure 4-10. ..................................................................................... 85
Figure 4-12: Transmission spectrum of the 4x1x1 array simulated by HFSS with
PEC boundary conditions. ....................................................................................... 86
Figure 4-13: Measured (within a waveguide) transmission spectrum of the 4x1x1
SRR array. ................................................................................................................ 87
xvii
Figure 4-14: Fabricated SRR unit cell and probes used to measure its transmission
spectrum. .................................................................................................................. 87
Figure 4-15: Comparison of the fabricated SRR unit cell and 4x1x1 SRR array. ... 88
Figure 4-16: A simple experimental set-up used to measure transmission spectrum
of SRR structures in air. ........................................................................................... 88
xviii
CHAPTER 1
INTRODUCTION
Research on metamaterials, which are engineered materials with unusual properties
not occurring naturally, has become increasingly important over the last decade.
The first and one of the most important contributions to this topic was made in 1968
by V.G. Veselago who showed that materials with both negative permittivity and
negative permeability are theoretically possible. Veselago used the term “lefthanded medium” for such materials as the E , H and k vectors of plane wave
propagation form a left-handed mutually perpendicular vector set instead of a righthanded one. In his paper, it is also indicated that in left-handed media wave vector
k and the Poynting vector S are in opposite directions. Because of the reversed
direction of k vector, phase velocity is also reversed in these materials. Veselago
implied that as a result of these reversals, Doppler Effect, Snell’s Law and
Cherenkov radiation will all be reversed in left-handed materials [1]. The next
important contribution was made almost 30 years later, in 1999, by Pendry et al.
They demonstrated experimentally that “Split-Rings” are useful to obtain negative
effective permeability µeff [2]. The following year, in 2000, Smith et al
experimentally demonstrated that the use of thin wire arrays in addition to SRR
(Split Ring Resonator) arrays provided negative effective permittivity, єeff, and
negative effective permeability, µeff, simultaneously over a common frequency band
[3].
The conventional split-ring resonator unit cell suggested by Pendry et al was
composed of two circular coplanar metallic rings each with a split displaced by 180
degrees. These rings were printed on a low-loss dielectric substrate having the same
center and separated from each other by a short gap distance as shown in Figure 1-1
1
(a). Bianisotropic behavior of this SRR unit cell structure was investigated in 2002
by Marques et al in [4]. A modified version of SRR (later called as broadside
coupled (BC) SRR) was suggested in the same paper to avoid bianisotropy. Next
year, in 2003, comparative analysis of the conventional (or edge-coupled) SRR and
BC-SRR was given in [5] where metallic rings of the BC-SRR were printed on both
sides of the dielectric substrate and aligned in such a way that their splits were
displaced by 180 degrees as shown in Figure 1-1 (b). SRR unit cells with multiple
rings were suggested and analyzed later by Bilotti et al in 2007 [6]. These and many
other studies on the theory and analysis of SRRs have appeared in the metamaterial
literature so far. In the mean time, SRRs have been used in diverse applications at
microwave and optical frequencies extending from superlenses [7, 8] to cloaking
[9].
Figure 1-1: Unit cell geometry for a) Conventional SRR, b) BC-SRR.
2
Most of the investigations on SRRs have involved either measurements or
numerical simulations of SRR unit cell or array topologies. Studies for developing
equivalent circuit models for SRR structures are still in their crawling phase. The
results obtained and reported in literature in this important area are neither complete
nor fully consistent yet.
Describing SRR structures by their equivalent circuit models is important for two
major reasons: First of all, transmission/reflection spectra obtained by
measurements or by highly complex commercial simulation tools do not provide
enough insight about the operational mechanism of the SRR topology under
investigation. When an equivalent circuit model is available, it is much easier to
establish explicit relationships between the physical properties (i.e. electrical
parameters, dimensions, etc.) of the SRR structure and its frequency dependent
transmission/reflection behavior. Secondly, use of equivalent circuit models makes
a computationally efficient optimization approach possible in metamaterial design.
Using commercial full-wave electromagnetic solver packages such as Ansoft HFSS
or CST Microwave Studio is not feasible in the analysis phase of an optimization
process since such solvers would have intolerably long run times at each iteration.
Instead, the analysis of a metamaterial structure can be completed in a fraction of a
second in each iteration of the optimization process if an equivalent circuit model is
available for the metamaterial topology to be designed. Hence, not only the
resonance frequency but also the overall S-parameter spectra of an SRR array can
be optimally estimated (or simply calculated without optimization) by using
sufficiently accurate equivalent circuit models which account for the loss effects
also.
In recent years, several researchers have contributed to the area of SRR modeling
[5, 6, 10-17]. In 2003, following the work done in [4], Marques et al proposed a
simple RLC equivalent circuit model for conventional SRRs in the presence of a
dielectric substrate [5]. In 2005 Baena et al analyzed SRR and complementary SRR
(CSRR) structures coupled to planar transmission lines and proposed equivalent
3
circuit models for the SRR and transmission line combinations. SRR unit cells are
represented by simple resonant LC circuits in their work [10]. In 2005, Qun Wu et
al [11] suggested an equivalent circuit model for a conventional SRR cell based on
the quasi-static approach. In 2006, Johnson et al analyzed two elements of SRR
array with co-planar and parallel plate capacitance approach [12]. In 2007, Bilotti et
al introduced multiple SRR structures and modeled them using as LC resonance
circuits. They also provided total inductance and total capacitance expressions [6].
In 2008, Wang et al modeled a modified SRR structure where the resonance
frequency of the structure could be changed by rotating the inner ring of the SRR
[13].
The main objective of this thesis is to establish useful equivalent circuit models for
SRR unit cell and array topologies. Extension of metamaterial applications from
microwave frequencies to terahertz (THz) and optical frequencies [18, 19] requires
geometrical simplicity as the wavelengths become very small (in the order of
submilimeters and microns) in the optical range and the SRR cells must be of
subwavelength size. For that reason, single ring SRR topologies are investigated in
this thesis. Square shaped single metallic rings with multiple gaps and their array
forms are analyzed, designed and modeled by lumped circuit elements. Use of twoport equivalent circuit representations of these elements is suggested in particular to
compute the complex S-parameters for SRR arrays. The transmission spectra (i.e.
magnitudes of S21 spectra) estimated by equivalent circuit models and computed by
simple Matlab codes are compared with transmission spectra computed via HFSS
simulations. The results are found to be in good agreement in most cases, although
the equivalent circuit models can describe the SRR array topologies only
approximately. For a selected square ring SRR structure resonating in 10-13 GHz
band, experimental results are also obtained. An SRR unit cell and a four-element
SRR array, extending in the propagation direction are manufactured and measured
within a metallic waveguide environment.
4
In Chapter 2, basics of numerical HFSS simulations and equivalent circuit modeling
for SRR structures are presented. Importance of choosing proper boundary
conditions and proper computational volume dimensions to simulate SRR behaviors
are discussed and demonstrated by examples. In this context, effect of periodicity
parameters on the resonance frequency is investigated. Basics of two-port
equivalent circuit modeling for SRR structures are also discussed in Chapter 2.
In Chapter 3, unit cell structure and arrays of single loop square shaped SRRs with
four splits are analyzed and modeled. These SRR structures are found to have
resonance frequencies around 35 GHz. Using two-port circuit representations and
Matlab programming, transmission spectra of these topologies are estimated.
Modeling results are compared with HFSS simulation results for validation.
In Chapter 4, a four split single square loop SRR unit cell and a four-element SRR
array (named as 4x1x1 array) are designed and fabricated. In addition to examining
these structures via HFSS simulations and equivalent circuit modeling, their
transmission spectra are also measured over the frequency band from 10 GHz to 13
GHz. All these results are found to be in good agreement.
Finally, conclusions of this thesis work along with possible future work suggestions
are given in Chapter 5.
5
CHAPTER 2
BASICS OF NUMERICAL SIMULATIONS AND
EQUIVALENT CIRCUIT MODELING FOR SRR
STRUCTURES
2.1
Introduction
In this chapter, basics of numerical simulations and equivalent circuit modeling for
SRR structures (unit cells or arrays) will be discussed. First, examples of HFSS
simulations will be presented with special emphasis on the choice of boundary
conditions and on the dimensions of computational volume. Then, basics of the
equivalent circuit modeling for SRR unit cells and arrays will be introduced. The
notation used for array forms is shown in Figure 2-1.
Figure 2-1: Notation used to identify SRR array topologies.
6
2.2
Use of HFSS Simulations to Compute the Transmission
Spectra of SRR Structures
Use of commercial full wave electromagnetic solvers such as Ansoft’s HFSS or
CST Microwave Studio is very common in the analysis of metamaterial structures.
In this thesis HFSS, which is based on finite elements method (FEM), is used for
the analysis of SRR structures. Proper use of boundary conditions is very important
in HFSS simulations. To simulate the behavior of a single SRR unit cell in
isolation, perfectly matched layer (PML) type of boundaries must be implemented.
PMLs are frequency dependent structures. While using these layers as boundaries,
it is vital to define the minimum reference frequency of the simulation frequency
range. The use of perfect electric conductor (PEC) or perfect magnetic conductor
(PMC) type boundary conditions around an SRR unit cell, on the other hand, leads
to the simulation of infinite SRR array structures due to the formation of images of
the SRR unit cell with respect to the PEC and PMC boundaries.
It is well known in metamaterial literature that SRR structures are magnetic
resonators providing negative permeability (µ-negative or MNG) regions around
their resonance frequencies. As to be discussed in this chapter, a single loop SRR
cell can be modeled by a simple RLC resonant circuit. The inductance “Lself” refers
to the self inductance of the metallic loop and the capacitance “C” is created at the
split location as a result of magnetic induction. The incident time-varying magnetic
field vector must be perpendicular to the SRR plane to excite the SRR’s magnetic
resonances. To compute the complex S-parameters of an infinite (in the E-field and
H-field directions) SRR array by HFSS under plane wave excitation, PEC and PMC
boundary conditions are frequently used in literature.
Location of the PEC and/or PMC boundaries (i.e. the dimensions of the
computational volume) is also important as their distance to metallic inclusions
determine the parameters of periodicity of the SRR array. Computed
transmission/reflection spectra, values of the resonance frequencies (i.e. the
7
complex S-parameter spectra, in general) change according to the type of boundary
conditions and location of the boundaries. Similar concerns regarding the location
of PMLs are also important in the simulation of isolated SRR cells. All these issues
will be addressed in the following subsections.
2.2.1 Simulation Problem 1
In this subsection, transmission spectrum of the SRR unit cell shown in Figure 2-2
(a) is simulated via HFSS over the frequency range 30-40 GHz by using PEC and
PMC boundary conditions. Dimensions of this SRR unit cell are as follows: Side
length of SRR (L) is 2.8 mm, gap width (g) and metal strip width (w) are both 0.3
mm, substrate dimensions (D=Dx=Dy) along the x and y directions are both 4 mm,
and substrate thickness (h) along the z direction is 0.5 mm. Gold is used for metal
inclusions and a low loss dielectric material with relative permittivity ( ε r ) of 4.6
and dielectric loss tangent ( tan α ) of 0.01 is used as the substrate. The PEC
boundary conditions are applied at those surfaces of the computational volume
which are perpendicular to the E field vector. Similarly, PMC type boundary
conditions are applied at those surfaces of the computational volume which are
perpendicular to the H field vector. The remaining two surfaces are obviously
labeled as the input and output planes in Figure 2-2 (b).
8
Figure 2-2: Schematic views of SRR unit cell and the HFSS simulation set-up a) Geometry and
dimensions of the SRR unit cell, b) Boundary conditions used for HFSS simulations.
Due to the imaging effects of PEC and PMC boundaries, the computed transmission
spectrum (i.e. S 21 versus frequency curve) belongs to a two dimensional infinite
SRR array as explained in Figure 2-3 which shows periodicity of the array in E
field and H field directions, respectively.
9
Figure 2-3: Two dimensional periodic SRR array implemented by the use of PEC and PMC
boundary conditions in HFSS simulation a) Array in E field direction, b) Array in H field
direction.
First, the transmission spectrum of this SRR array is simulated by using the
dimension DH =1.75 mm where DH is the depth of the vacuum regions over and
under the SRR cell within the computation volume in the direction of H field as
seen in Figure 2-3 (b). The resultant transmission curve is plotted in Figure 2-4 (the
red one) with a resonance frequency very close to 34 GHz. Next, the simulation is
repeated by keeping all the parameters the same but doubling the value of the
parameter DH this time. The transmission spectrum of the SRR array with DH =3.5
mm is plotted in Figure 2-4 (the blue curve), indicating a shift of approximately 0.3
GHz in resonance frequency which is now 34.28 GHz. The periodicity distance Dz
10
in H field direction (along the z axis) is given by Dz = 2 DH + h where h=0.5 mm
is the substrate thickness. The periodicity distance Dz is changed from 4 mm to 7.5
mm in these simulations resulting a negligible shift of only 0.3 GHz, which
corresponds to a (0.3/34)x100=0.88 % change, even less than one percent. This is
an expected observation as the array in z direction is relatively sparse leading to the
result that the inductive coupling effects between array elements are not strong
along this dimension of the SRR array. It should be also emphasized that there is no
capacitive coupling between the SRR elements stacked along the H field direction
of the array. Due to the PMC boundary condition, the image of the SRR unit cell is
simulated to form an array in H field direction but charges accumulated at the split
locations of an image cell have exactly the same polarity pattern as that of the
original SRR cell.
Next, similar simulations are performed by changing the distance DE between the
PEC boundary and the outer border of the metallic strip, which determines the
periodicity distance Dy = 2 DE + L along the E field direction. Results will be
reported in the next subsection.
11
Figure 2-4: Effect of vacuum length in H field direction on resonance frequency (red curve
DH =1.75 mm, blue curve DH =3.5 mm).
2.2.2 Simulation Problem 2
In this simulation problem, variation of the resonance frequency of the SRR array in
response to changing the periodicity distance along the E -field direction will be
investigated by changing length ( DE ) of the dielectric region from the metal strip to
the PEC boundary. The same square shaped SRR unit cell of the first simulation
problem is also used in this simulation. The SRR array is simulated for three
different values of the parameter DE = 0.6, 0.9 and 1.2 mm corresponding to the
periodicity distance of D y = 4, 4.6 and 5.2 mm along the y-axis. Resulting
transmission spectra are plotted in Figure 2-5 below.
12
Figure 2-5: Variation of transmission minimum with substrate length in electric field direction.
As observed in this figure, resonance frequency of the SRR array decreases from
33.96 to 31.58 GHz as the periodicity distance along the electric field direction
increases from 4 mm to 5.2 mm, which is an expected result. Because, due to the
perfect electric conductor (PEC) boundary shown in Figure 2-3 (a), the original
SRR cell and its mirror images carry induced currents in the same direction.
Therefore, a subtractive mutual inductance is created between the adjacent SRR
loops. As the parameter DE gets larger, mutual inductance (M) obviously gets
smaller, leading to increased equivalent inductance (Leq) values. In the mean time,
capacitive coupling between neighboring SRR cells also diminishes with increased
separation distance. Decreasing coupling capacitance Cmutual will cause a smaller
equivalent capacitance (Ceq) as the equivalent gap capacitance of each SRR cell and
the coupling capacitance between adjacent cells are in series. However, based on
the formulas given in [20], the coplanar component of the coupling capacitance,
which is dominant as compared to the parallel plate component, decreases by the
natural logarithm of the increasing separation distance between the adjacent SRR
13
elements while the mutual inductance term decreases almost linearly under the
same effect. Therefore, the increase in Leq is found to be faster than the decrease in
Ceq and the product Leq Ceq increases causing the resonance frequency
ω0 = ( Leq Ceq ) −1 / 2 to decrease as D E gets larger. To demonstrate this discussion
graphically, a MATLAB code is written to compute the mutual inductance (M) and
coupling capacitance (Cmutual) for various values of the separation distance d=2 D E
between the SRR cells in y-direction changing from 1.2 mm to 2.4 mm with 0.2
mm steps. Resulting curves are plotted in Figure 2-6 and Figure 2-7.
-10
6
x 10
Mutual Inductance (Henry)
5.5
5
4.5
4
3.5
3
1
1.2
1.4
1.6
1.8
2
2.2
2.4
d (mm)
Figure 2-6: Variation of mutual inductance with cell to cell separation distance along the
electric field direction.
14
-14
3.4
x 10
Mutual Capacitance (Farad)
3.3
3.2
3.1
3
2.9
2.8
2.7
1
1.2
1.4
1.6
1.8
2
2.2
2.4
d (mm.)
Figure 2-7: Variation of mutual capacitance with cell to cell separation distance along the
electric field direction.
Rate of decreases of the mutual inductance and mutual capacitance curves (i.e. their
local slopes at the sampling points) are also computed and plotted against d=2 DE
in Figure 2-8 and Figure 2-9. It is indeed seen that mutual inductance decreases
more sharply than the mutual capacitance. Finally, the resonance frequency f 0 of
the array is plotted versus d=2 DE in Figure 2-10 showing the expected decrease in
f 0 with increasing separation distance d based on the calculated circuit parameters.
15
1.5
Rate of Change of Mutual Inductance
1
0.5
0
-0.5
-1
-1.5
-2
-2.5
-3
1
1.2
1.4
1.6
1.8
2
2.2
2.4
d (mm.)
Figure 2-8: Rate of change of mutual inductance with d.
1.2
Rate of Change of Mutual Capacitance
1
0.8
0.6
0.4
0.2
0
-0.2
-0.4
-0.6
-0.8
1
1.2
1.4
1.6
1.8
d (mm.)
2
2.2
Figure 2-9: Rate of change of mutual capacitance with d.
16
2.4
36.4
36.2
36
Frequency (GHz)
35.8
35.6
35.4
35.2
35
34.8
34.6
1
1.2
1.4
1.6
1.8
2
2.2
2.4
d (mm.)
Figure 2-10: Variation of resonance frequency with d based on calculated circuit parameters.
2.2.3 Simulation Problem 3
As discussed at the beginning of section 2.2, type of boundary conditions (whether
PEC/PMC or PML) changes the nature of the SRR simulation problem drastically
and produces different transmission spectrum curves. To observe the effect of
mutual interactions between neighboring SRR cells within an array topology, we
have simulated four different array structures of dimensions 1x2x1, 1x3x1, 1x4x1
and 1x5x1 (i.e. one dimensional arrays of SRR cells in E-field direction) in addition
to a single SRR unit cell by using HFSS software with PML boundary conditions.
The metallic strips of the SRR unit cell are made of copper with 0.035 mm
thickness and the 0.762 mm thick substrate is made of the AD350 dielectric
material with the relative permittivity of єr=3.5. Side length of the square shaped
SRR loop (L) is 8 mm, gap width (g) is 0.3 mm and metal strip width (w) is 0.6
17
mm. The dielectric substrate has dimensions of Dx=Dy=10 mm both in x and y
directions within one period of the array (see Figure 2-11). Transmission spectra of
the infinite SRR array (implemented by PEC boundary conditions) and that of the
isolated SRR cell (simulated by using the PML boundary conditions) were quite
different from each other, as shown in Figure 2-12. However, transmission
spectrum curves of the isolated “1x2x1 array” and isolated “1x3x1 array” looked
more and more similar to that of the infinite SRR array, as expected, due to the
included mutual interaction effects. Note that “isolation” condition is satisfied by
using PML boundary conditions in HFSS simulations. Obviously, the isolated
“1xnx1 array” should behave more and more similarly to the infinite array as n
(number of array elements in E-field direction) gets larger. To see further
convergence to the transmission spectra of the infinite array, the isolated “1x4x1
array” and “1x5x1 array” were also simulated. However, for these last two HFSS
simulations the expected “converging” transmission spectrum curves could not be
obtained (and the HFSS simulations needed extremely long run-times) due to
insufficient mesh numbers (see Figure 2-12 and Table 2-1) used in FEM
computations.
Figure 2-11: Unit cell geometry used for comparison of boundary conditions.
18
Figure 2-12: Comparison of transmission spectrum of different SRR configurations with PEC
and PML boundaries using DE=1 mm and DH=11.049 mm.
Table 2-1: Comparison of mesh numbers used in HFSS simulations with PML boundary
conditions for different number of array elements in electric field direction
Number
of SRR
elements
Mesh
Number
Single
SRR
1x2x1
Array
1x3x1
Array
1x4x1
Array
1x5x1
Array
13711
19675
23425
24621
30522
As seen in Table 2-1, number of elements of the 1x4x1 array and hence its
computational volume is doubled as compared to those for the 1x2x1 array, for
instance, but the mesh number is increased by only 25 percent. If these simulations
were run using a more powerful computer, much better numerical resolution would
be obtained as a result of using sufficiently large mesh numbers. Therefore the
transmission curves to be obtained for the 1x4x1 and 1x5x1 arrays would look
much more similar to 1x ∞ x1 array result shown in Figure 2-12 which is obtained
19
when the transmission spectra of an SRR unit cell is simulated by HFSS with PEC
boundary conditions.
So far, issues relevant to HFSS simulations of SRR structures have been discussed.
In the following section, basics of equivalent circuit modeling for SRR unit cells
and arrays will be presented.
2.3
Basics of SRR Modeling by Equivalent Lumped Circuit
Models
The need for developing equivalent circuit models for SRR structures has already
been discussed in the Introduction chapter. Most of the studies reported in literature
[5, 6, 10-12,16] have attempted to represent the conventional SRR or BC-SRR unit
cell structures by a simple LC resonant circuit so that the resonance frequency can
be obtained from the well-known formula f 0 = (2π LC ) −1 where C is the
equivalent lumped capacitance and L is the equivalent lumped inductance of this
basic model. A few of these papers included the loss effects into the SRR cell
model by suggesting equivalent resistance expressions as well [5, 6, 11].
Expressions of L, C and R are functions of the SRR unit cell geometry, dimensions
and electrical parameters of the metal and dielectric substrate materials. The mutual
capacitance and mutual inductance values pertinent to the conventional two ring
SRR and multiple-ring SRR [6] unit cells are implicitly included in the equivalent
capacitance and inductance expressions used in the above mentioned references.
Johnson et al [12] provided the expressions for the mutual inductance and coupling
capacitance effects between two neighboring (single ring with single split type)
SRR unit cells resonating at infrared frequencies. The main purpose of equivalent
circuit modeling was the estimation of resonance frequency in all the studies
mentioned above except for the work in [11] where the estimation of the Sparameter spectra was demonstrated by Wu et al.
20
In this thesis, not only the resonance frequency values but also the transmission
spectrum of SRR unit cells (single square ring with multiple split type) and SRR
arrays are estimated by using two-port equivalent circuits to model individual SRR
behaviors and coupling effects resulting from array topologies. Ohmic loss effects
(taking place in conductors and dielectric parts) are also taken into account to
improve the accuracy of the equivalent circuit models. Then, transmission spectra
of the investigated SRR structures are computed with simple and fast Matlab codes
by using Z-parameter, Y-parameter and chain parameter matrices, whenever
needed. The results are converted finally to S-parameter forms since the
transmission spectrum is the magnitude of the complex S21 spectrum of the overall
SRR array. Accuracy of the resulting equivalent circuit models is tested by
comparing the modeling results with the results obtained from HFSS simulations.
2.3.1 Modeling an SRR Unit Cell by a Proper RLC Circuit
There have been attempts in literature to model individual SRR unit cells by LC
circuits and compute the resonance frequency from f 0 = (2π LC ) −1 formula.
Since our purpose is to compute the whole transmission spectrum (or the complex
S-parameters S21, S11, etc. in general), we should do more than that and suggest a
properly described two-port circuit representation for the SRR unit cell considered.
At this point, four possibilities have emerged: (i) Series RLC resonant circuit in the
shunt branch, (ii) Parallel RLC resonant circuit in the shunt branch, (iii) Series RLC
resonant circuit in the series branch or (iv) Parallel RLC resonant circuit in the
series branch of the two-port representation of the SRR. The feasible or most
suitable resonant two-port configuration is determined as outlined below. For
simplicity, but without any loss of generality, the loss effects are neglected in the
following investigation.
Case 1: If the SRR unit cell is represented by a resonant LC circuit as in the cases
(i) or (ii) mentioned above, the corresponding equivalent two-port circuit
21
representations will look like the one shown in Figure 2-13 where Z is the
equivalent impedance of the lumped elements in the shunt branch.
I2
I1
+
+
V1
V2
Z=1/Y
-
-
Figure 2-13: Two-port equivalent circuit representation of a single loop SRR cell with either
series or parallel LC resonant circuit in the shunt branch.
The Z-matrix representation of this two-port circuit is given as:
⎡V1 ⎤ ⎡ Z 11
⎢V ⎥ = ⎢ Z
⎣ 2 ⎦ ⎣ 21
Z 12 ⎤
Z 22 ⎥⎦
Z11=
V1
I1
I 2 =0
Z21=
V2
I1
I 2 =0
Z12=
V1
I2
I1 =0
Z22=
V2
I2
I1 =0
⎡ I1 ⎤
⎢I ⎥
⎣ 2⎦
(1)
=Z
(2)
=Z
(3)
=Z
(4)
=Z
(5)
22
Hence the [Z ] matrix is obtained as:
⎡Z
[Z ] = ⎢Z
⎣
Z ⎤
Z ⎥⎦
(6)
The Y-matrix representation of the same two-port is given by:
⎡ I 1 ⎤ ⎡Y11 Y12 ⎤ ⎡V1 ⎤
⎢ I ⎥ = ⎢Y
⎥⎢ ⎥
⎣ 2 ⎦ ⎣ 21 Y22 ⎦ ⎣V2 ⎦
(7)
However, the [Y ] matrix is not defined for this particular two-port circuit as
[Y ] = [Z ] -1 and det(Z)= Z =0
As another alternative, chain [ABCD] matrix representation of this two-port circuit
can be obtained from:
⎡V1 ⎤ ⎡ A B ⎤ ⎡ V2 ⎤
⎢ I ⎥ = ⎢C D ⎥ ⎢− I ⎥
⎦ ⎣ 2⎦
⎣ 1⎦ ⎣
V1
V2
(8)
=1
(9)
B=
V1
=0
− I 2 V2 =0
(10)
C=
I1
V2
(11)
A=
D=
I 2 =0
=Y
I 2 =0
I1
=1
− I 2 V2 =0
⎡A B⎤ ⎡ 1
⎢C D ⎥ = ⎢Y
⎣
⎦ ⎣
23
0⎤
1⎥⎦
(12)
(13)
This representation will be used later in Chapters 3 and 4 for model computations.
It should be noted that the conversion between impedance parameters and chain
parameters is given by the rule [21]:
⎡ Z11
⎡ A B ⎤ ⎢ Z 21
⎢C D ⎥ = ⎢ 1
⎣
⎦ ⎢
⎢⎣ Z 21
ΔZ ⎤
Z 21 ⎥ = ⎡ 1
⎥
Z 22 ⎥ ⎢⎣Y z
Z 21 ⎥⎦
0⎤
1⎥⎦
(14)
where
ΔZ =det(Z)
(15)
It is also possible to obtain the scattering parameters of a given two-port network
from its Z-parameters as summarized below [21]:
2
( Z 11 − Z 0 )( Z 22 + Z 0 ) − Z 12 Z 21 ( Z − Z 0 )( Z + Z 0 ) − Z
=
S11=
( Z 11 + Z 0 )( Z 22 + Z 0 ) − Z 12 Z 21 ( Z + Z 0 )( Z + Z 0 ) − Z 2
S11=
S21=
(16)
2 Z 21 Z 0
2ZZ 0
=
( Z 11 + Z 0 )( Z 22 + Z 0 ) − Z 12 Z 21 ( Z + Z 0 )( Z + Z 0 ) − Z 2
S21=
S12=
− Z0
2Z + Z 0
2Z
2Z + Z 0
(17)
2ZZ 0
2 Z 12 Z 0
=
( Z 11 + Z 0 )( Z 22 + Z 0 ) − Z 12 Z 21 ( Z + Z 0 )( Z + Z 0 ) − Z 2
S12=
2Z
=S21
2Z + Z 0
2
( Z 11 − Z 0 )( Z 22 + Z 0 ) − Z 12 Z 21 ( Z − Z 0 )( Z + Z 0 ) − Z
S22=
=
( Z 11 + Z 0 )( Z 22 + Z 0 ) − Z 12 Z 21 ( Z + Z 0 )( Z + Z 0 ) − Z 2
24
(18)
S22=
− Z0
=S11
2Z + Z 0
(19)
The parameter Z0 used in equations (16)-(19) is the terminal impedance at the input
and output ports, by definition. Data for this normalization factor is provided by the
HFSS simulation of the analyzed SRR structure while comparing the HFSS
simulation results and the equivalent circuit modeling results. As mentioned earlier,
we have two possibilities for the topology of the shunt branch in Figure 2-13, a
series LC resonant circuit as shown in Figure 2-14 (a) or a parallel LC resonant
circuit as shown in Figure 2-14 (b). In both cases, the structure resonates at
2πf 0 = ω 0 =
1
LC
.
Figure 2-14: Two different representations for the impedance Z a) Series LC circuit, b)
Parallel LC circuit.
For the first case in Figure 2-14 (a), Z becomes:
Z = jωL +
1 − ω 2 LC
1
=
j ωC
jωC
25
(20)
Obviously, we have Z=0 at the resonance frequency f 0 =
1
2π LC
. Also, from
equation (17), it can be seen that the transmission spectrum becomes also zero as
S21=0 at f= f 0 . It can be also concluded from equations (17) and (20) that S21
approaches to unity as frequency f approaches to both zero and infinity. In other
words, the two-port circuit topology shown in Figure 2-14 (a) has the expected
“band-stop” or “notch” type transmission spectrum with a minimum (at the zero
level in this lossless case) at the resonance frequency f 0 .
For the second case shown in Figure 2-14 (b), on the other hand, the impedance Z
becomes:
Z=
1
jωC +
1
jω L
=
jωL
1 − ω 2 LC
(21)
Obviously, Z approaches to infinity at the resonance frequency f= f 0 . Therefore,
S21 approaches to unity at resonance based on the equation (17) which is not an
expected behavior for the transmission spectrum of the SRR unit cell. In
conclusion, parallel LC resonant circuit in the shunt branch of Figure 2-14 (b) is not
an acceptable circuit representation for the SRR while the series LC resonant circuit
in the shunt branch is a perfectly acceptable choice. Figure 2-15 shows an improved
version of this acceptable equivalent circuit model where the total conductor losses
are represented by the resistance, Rc, which is connected in series with the total self
inductance L, and dielectric losses around the gaps are represented by the
resistance, Rd, which is connected in parallel to the equivalent gap capacitance, C.
26
Figure 2-15: A feasible two-port equivalent circuit representation for SRR unit cell including
ohmic loss effects.
The impedance Z for this improved branch model is given as:
⎛ 1
Rd
⎜
jωC
⎜
Z = R C + j ωL +
⎜ 1
+ Rd
⎜
ω
j
C
⎝
⎞
⎟
⎟
⎟
⎟
⎠
(22)
Inserting Z into the equations (16) and (17), S-parameters S11 and S21 can be
obtained as:
S11=
S21=
− (1 + jωCR d ) Z 0
2 RC + 2 Rd − 2ω LCR d + jω (2CRC Rd + 2 L + Z 0 CR d ) + Z 0
2
(23)
2 RC + 2 Rd − 2ω 2 LCR d + 2 jω (CRC Rd + L)
2 RC + 2 Rd − 2ω 2 LCR d + jω (2CRC Rd + 2 L + Z 0 CRd ) + Z 0
(24)
4l
σδ 2( w + t )
(25)
g
(tan α )(ωε ) wh
(26)
Rc =
Rd =
27
The parameters used in equation (25) are defined as follows: l is the side length of
SRR, σ is the conductivity of metal, δ is skin depth in metal, w is width of the metal
stripes and t is the thickness of metal layer. While modeling the conductor losses by
Rc , good conductor assumption has been made. The perimeter of the cross section
of square SRR is given by 2(w + t) and multiplying this expression by skin depth
gives the effective current carrying surface over the cross section of metallic strip.
For good conductors, skin depth δ is given as
δ=
1
πfμσ
(27)
In the expression of Rd , on the other hand, conductivity σ d of the low-loss (good
dielectric) substrate is replaced by σ d = ω ε tan α where ω = 2π f is the angular
frequency, ε = ε 0 ε r is the permittivity of the substrate and tan α is the loss tangent
of the substrate. As defined earlier, w is the width of metal strip and h is the
thickness of substrate. As the substrate thickness is small enough (only 0.5 mm) for
the present SRR geometry, electromagnetic waves in the gap (split) region are
assumed to penetrate into the substrate through its whole depth along the z
direction.
A Matlab code is written to compute the transmission spectrum of SRR unit cell
using Equations (24) through (27) for the SRR parameters used in section 2.2.3.
The resulting S 21 versus frequency curve is plotted in Figure 2-16 together with the
transmission spectrum of this SRR obtained from the HFSS simulation with PML
boundary conditions. In this simulation, separation distance DH is taken to be 5 mm
between the SRR surface and the PML layer. These two curves are found in good
agreement.
28
Figure 2-16: Transmission spectrum curves of the isolated SRR cell, which are obtained by
HFSS simulation with PML boundary conditions (with DE=1 mm and DH=5 mm) and by
equivalent circuit modeling approach. Geometry and dimensions for the SRR cell are
described in Section 2.2.3. Equivalent circuit model is given in Figure 2-15.
As seen in Figure 2-16, the resonance frequency obtained from the equivalent
circuit model is 11.89 GHz while the HFSS simulation result gives 11.90 GHz.
Also, the overall shapes of these two waveforms of transmission spectrum are
reasonably close to each other over the whole computational frequency bandwidth.
Case 2: To model a single loop SRR with a two port equivalent circuit model, there
are two other possibilities; a series LC resonant circuit or a parallel LC resonant
circuit placed in the series branch of the two-port. Possibility of both cases can be
investigated using the two-port circuit shown in Figure 2-17.
29
Figure 2-17: Alternative two-port equivalent circuit representation of a single loop SRR with
either series or parallel LC resonant circuit in the series branch.
The Y-matrix representation of this two-port is given as:
⎡ I 1 ⎤ ⎡Y11 Y12 ⎤ ⎡V1 ⎤
⎢ I ⎥ = ⎢Y
⎥⎢ ⎥
⎣ 2 ⎦ ⎣ 21 Y22 ⎦ ⎣V2 ⎦
(28)
where
Y11 =
Y21 =
Y12 =
I1
V1 V
I2
V1 V
I2
V1 V
Y22 =
=Y
(29)
= −Y
(30)
= −Y
(31)
2 =0
2 =0
2 =0
I2
=Y
V2 V =0
(32)
−Y⎤
Y ⎥⎦
(33)
1
[Y ] = ⎡⎢
Y
⎣− Y
−1
The Z-matrix representation of this two-port is not defined as [Z ] = [Y ] where
determinant of the [Y ] matrix is zero.
30
Also, the chain parameter representation of this two-port can be obtained as
⎡ V2 ⎤
⎢− I ⎥
⎣ 2⎦
(34)
=1
(35)
=Z
(36)
⎡V1 ⎤ ⎡ A B ⎤
⎢ I ⎥ = ⎢C D ⎥
⎦
⎣ 1⎦ ⎣
A=
B=
I 2 =0
V1
− I2
C=
D=
V1
V2
I1
V2
V2 =0
=0
(37)
I 2 =0
I1
− I2
=1
(38)
V2 =0
⎡ A B ⎤ ⎡1 Z ⎤
⎢C D ⎥ = ⎢0 1 ⎥
⎣
⎦ ⎣
⎦
(39)
Conversion between the Y-parameters and the chain [ABCD] parameters is possible
according to the following rules [21]:
⎡ Y22
−
⎡ A B ⎤ ⎢ Y21
⎢C D ⎥ = ⎢ ΔY
⎣
⎦ ⎢−
⎢⎣ Y21
1 ⎤
⎡
Y21 ⎥ ⎢ 1
⎥=
Y
⎢
− 11 ⎥ ⎣⎢ 0
Y21 ⎥⎦
−
ΔY = [Y ] = det ([Y ])
1
Y
1
⎤
⎥
⎥
⎦⎥
(40)
(41)
Finally, the S-Parameters can be obtained from the Y-Parameters [21] as described
below:
31
(Y − Y11 )(Y0 + Y22 ) + Y12 Y21 (Y0 − Y )(Y0 + Y ) + Y
S11= 0
=
(Y11 + Y0 )(Y22 + Y0 ) − Y12 Y21 (Y0 + Y )(Y0 + Y ) − Y
S11=
S21=
2
2
Y0
=S22
2Y + Y0
(42)
− 2Y21Y0
2YY0
=
(Y11 + Y0 )(Y22 + Y0 ) − Y12 Y21 (Y + Y0 )((Y + Y0 ) − Y 2
S21=
2Y
=S12
2Y + Y0
where the normalization term Y0 = 1
(43)
is the terminal admittance seen at the
Z0
input/output ports, by definition. The magnitude of the S21 spectrum must result in a
transmission curve which is expected to be zero (in the lossless case) at the
resonance frequency f 0 =
1
2π LC
for the SRR unit cell under consideration. Also,
as seen from equation (43), Y=0 condition must hold at resonance to satisfy this
expectation. This point should be taken into account while evaluating the
possibilities of alternative two-port equivalent circuit topologies shown in Figure
2-18.
For the equivalent circuit shown in Figure 2-18 (a), the admittance Y is obtained as
given in equation (44) and it is seen that Y approaches to infinity at f 0 =
causing S 21
1
2π LC
to approach to unity (see equation (43)). As zero transmittance is
expected at resonance, this equivalent circuit can not be a feasible choice to
represent the SRR unit cell.
32
Figure 2-18: Alternative representations of two-port network consisting of series impedance
a) Series LC circuit, b) Parallel LC circuit.
Y=
jω C
1
=
Z 1 − ω 2 LC
(44)
For the two-port equivalent circuit model shown Figure 2-18 (b), on the other hand,
expression for the admittance Y is given in equation (45) and it becomes zero at the
resonance frequency making the transmittance also zero, as expected.
Y=
1 1 − ω 2 LC
=
Z
jωL
(45)
Therefore, this potentially useful equivalent circuit is investigated in more detail by
including the loss effects into the model as shown in Figure 2-19 where the
resistance Rc represents the conductor losses and the other resistance Rd represents
the dielectric losses occurring around the gap locations. The admittance expression
for this improved topology is given in equation (46).
33
Figure 2-19: Improved version of the two-port circuit representation shown in Figure 2-18 (b)
with the inclusion of equivalent loss resistances.
Y=
1 ( jωL + Rc )(1 + jωCR d ) + R d
=
Z
( jωL + Rc ) R d
(46)
Next, expressions for the scattering parameters S21 and S11 can be computed using
equations (42), (43) and (46) to get:
S11=
S21=
Y0 ( jωLR d + Rc Rd )
2 jωL − 2ω 2 LCR d + 2 jωCRc Rd + 2 Rc + 2 R d + Y0 ( jωLR d + Rc R d )
− 2( jωL − ω 2 LCR d + jωCRc Rd + Rc + Rd )
2 jωL − 2ω 2 LCR d + 2 jωCRc Rd + 2 Rc + 2 Rd + Y0 ( jωLR d + Rc R d )
(47)
(48)
Where the loss resistances Rc and Rd are computed from equations (25) and (26).
Again a Matlab code is developed to compute the S11 and S21 parameters. The
magnitude of S21 spectrum is plotted against frequency in Figure 2-20. The S 21
spectrum obtained by an HFSS simulation is also plotted on the same figure for
comparison. Although these two curves are found similar, the agreement between
the results obtained by HFSS and equivalent circuit models are not found good
enough as compared to the case demonstrated earlier in Figure 2-16.
34
Figure 2-20: Transmission spectrum curves of the isolated SRR cell, which are obtained by
HFSS simulation with PML boundary conditions (with DE=1 mm and DH=5 mm) and by
equivalent circuit modeling approach. Geometry and dimensions for the SRR cell is described
in Section 2.2.3. Equivalent circuit model is given in Figure 2-19.
Based on the conclusions derived from the results of section 2.3, the two-port
equivalent circuit model shown in Figure 2-15, i.e the series resonant RLC circuit in
the shunt branch of the two-port circuit, is found to be the best lumped circuit
model to represent the single loop SRR unit cell structure. Only this equivalent
circuit topology will be used in the remaining chapters of this thesis to develop
equivalent circuit models of SRR arrays.
35
CHAPTER 3
NUMERICAL SIMULATIONS AND TWO PORT
EQUIVALENT CIRCUIT MODELING OF SQUARE
SHAPED SINGLE LOOP SRR STRUCTURES
3.1
Introduction
In this chapter, unit cell and array topologies with square-shaped single loop foursplit SRR structures will be modeled using the novel two-port equivalent circuit
modeling approach presented in section 2.3 of the previous chapter. Individual SRR
unit cells of the arrays will be represented by the series RLC resonant circuit model
shown in Figure 2-15. Interaction effects (i.e. the inductive and capacitive coupling
effects between the array elements) and additional ohmic losses stemming from the
non-zero conductivity of the dielectric substrate will also be modeled using proper
two-port circuits.
Transmission spectrum of a given array structure will be computed to be the
magnitude of the S21 spectrum of the two-port circuit representation of the overall
SRR array. Z-parameter, Y-parameter and chain (ABCD) parameter representations
will be used whenever needed to obtain the S-parameter matrix of the given
topology. Results of equivalent circuit models will be compared to HFSS
simulation results for validation. Isolated SRR structures will be simulated using
PML type boundary conditions as shown in Figure 3-1 (a). Infinite SRR arrays, on
the other hand, will be simulated by using PEC/PMC type boundary conditions as
seen in Figure 3-1 (b).
36
Figure 3-1: Boundary conditions for HFSS simulations: a) PML boundary conditions to
examine an isolated SRR unit cell, b) PEC/PMC boundary conditions to examine a two
dimensional infinite SRR array.
In this chapter, in addition to the SRR unit cell, arrays of SRR cells with sizes
1x2x1, 2x1x1 and 2x2x1 will be analyzed. The notation used to identify SRR array
sizes is previously given in Figure 2-1.
3.2
Single-Loop Square-Shaped SRR Unit Cell
In the first subsection to follow, transmission spectrum of a single-loop square
shaped SRR unit cell with four identical gaps will be estimated by using both
equivalent circuit modeling approach and HFSS simulation approach. Then, in the
next subsection, an infinite SRR array structure will be examined.
3.2.1 Numerical Simulation and Equivalent Circuit Modeling of the
Isolated SRR Unit Cell
The isolated SRR unit cell investigated in this subsection has four gaps (splits) of
equal width (g) placed at the middle of each side of its square loop as shown in
37
Figure 3-2 (a) and (b). The copper strips forming this fully-symmetrical multi-split
ring are printed on a low-loss dielectric substrate with ε r =4.6 and tan α =0.01.
Dimensions of this SRR unit cell are given in Table 3-1. Figure 3-2 (b) shows the
instantaneous charge polarities induced across the gap locations, due to the
magnetic induction phenomena caused by a time-varying incident magnetic field as
discusses previously. Geometrical parameters of SRR cells are also indicated in
Figure 3-2 (b). The equivalent lumped circuit used to model an individual SRR cell
is given in Figure 3-2 (c) where the equivalent capacitance (Ceq) is equal to one
fourth of individual gap capacitance (Cgap) as all four of these equal gap
capacitances are connected in series. Computation of circuit parameters Lself (total
self inductance of the metal ring), Cgap, Rc (equivalent resistance representing
conductor losses) and Rd (equivalent resistance representing dielectric losses) will
be discussed later in this subsection.
38
Figure 3-2: a) Unit cell geometry of a single square loop SRR, b) Parameters of the isolated
SRR cell, c) Equivalent two-port circuit model of the SRR cell.
Table 3-1: Geometrical parameters of the SRR unit cell shown in Figure 3-2
Substrate
L (Side length
g (Gap
Dimensions
of the SRR)
Width)
Dx=Dy=4mm
h=0.5 mm
2.8 mm
0.3 mm
w (Width
t(Thickness
of Metal
of Metal
Strip)
Strip)
0.3 mm
1 µm.
To calculate the gap capacitance, both coplanar (Ccp) and parallel-plate (Cpp)
capacitance contributions should be taken into account [12, 20].These capacitance
39
contributions are calculated in this thesis by using equations (49) through (51).
Also, the total self inductance of the metal loop is calculated using equation (52) as
discussed in reference [22].
Ccp =
(ε r + 1)ε 0 ⎡ 1 + k ' ⎤
In ⎢2
⎥ F/m
2π
⎢⎣ 1 − k ' ⎥⎦
⎛ ⎛ p ⎞2 ⎞
⎟ ⎟
k = ⎜1 − ⎜⎜
⎜ ⎝ p + 2q ⎟⎠ ⎟
⎝
⎠
'
Cpp = ε
Lself ≈
wt
g
(49)
(50)
(51)
2μ 0 L ⎡
⎛ L ⎞ ⎤
sinh −1 ⎜
⎟ − 1⎥
⎢
π ⎣
⎝ w/ 2⎠ ⎦
(52)
Table 3-2: Calculated values of circuit parameters for the SRR unit cell
Ccp (49)
17.01*10-15 F
Cpp (51)
0.0407*10-15 F
Cgap =Cpp+Ccp
17.0507*10-15 F
Ctotal= Cgap/4
4.263 * 10-15 F
Lself (52)
58.71* 10-10 H
While calculating coplanar gap capacitance using equation (49) p is taken as the
gap width (g) and q is taken to be equal to (L-g)/2. Values of Ctotal and Lself are
independent of frequency and computed values for these circuit parameters are
listed in Table 3-2. Values of equivalent loss resistances Rc and Rd, however, are
functions of frequency. For that reason, they are not listed in Table 3-2. Typical
values of Rc and Rd are in the order of 1 Ω and 92 kΩ around the resonance
frequency.
40
Using all these circuit parameters, the transmission spectrum (i.e. S 21 versus
frequency curve) of this unit cell is computed by equations (24) through (27) as
described in section 2.3. A simple Matlab code with a very short run time is
developed for this purpose. The resulting transmission curve is plotted in Figure 3-3
over the frequency range 15-40 GHz. Transmission spectrum of the isolated SRR
cell is also computed by HFSS using PML boundary conditions as described in
Figure 3-1 (a). The PML boundaries are placed DH =2 mm above the SRR plane in
this HFSS simulation. In general, HFSS simulation results are found to be very
sensitive to the distance between the SRR plane and the PML boundary along the
H filed direction. The transmission spectrum computed by HFSS for DH =1 mm
separation distance between the SRR plane and PML boundary is also plotted in
Figure 3-4. Resonance frequency of the infinite SRR array is known to be around
34 GHz and the resonance frequency of the SRR unit cell is expected to be close to
this value. Therefore, the transmission spectrum simulated for the isolated SRR cell
for DH =1 mm case looks more reasonable as it has a resonance frequency around
f 0 =34 GHz. The normalization impedance Z0 is calculated as an output of the
HFSS simulations. As experience has shown, Z0 values change only slightly with
frequency when PEC/PMC boundary conditions are used to simulate SRR array
structures. On the other hand, Z0 becomes strongly dependent on frequency when
PML boundary conditions are implemented to simulate the isolated SRR unit cell
behavior. In order to compare the transmission spectrum results obtained by HFSS
and equivalent circuit modeling approaches, we need to convert Z-parameters of the
two-port model to S-parameters as described in Chapter 2. At this step, the
normalization impedance (Z0) values must be used in equation (17) at each sample
frequency. The Z0 output array provided by HFSS simulation is used for this
parameter conversion. Values of Z0 arrays provided in DH=2 mm and DH=1 mm
cases are appreciably different from each other. Therefore, the transmission
spectrum curve estimated by the two-port model is also affected by the use of this
normalization impedance data as shown in Figure 3-3 and Figure 3-4. Variations in
41
the transmission curves below 30 GHz, in particular, are thought to be spurious
effects caused by the use of PML boundary conditions.
Figure 3-3: Estimated transmission spectra of the SRR unit cell using HFSS simulations with
PML boundary conditions and DH=2 mm and using the equivalent two-port circuit given in
Figure 3-2 (c).
42
Figure 3-4: Estimated transmission spectra of the SRR unit cell using HFSS simulations with
PML boundary conditions and DH=1 mm and using the equivalent two-port circuit given in
Figure 3-2 (c).
3.2.2 Numerical Simulation and Equivalent Circuit Modeling of the
Infinite SRR Array along the Electric Field Direction
As discussed in Chapter 2 earlier, PEC/PMC type boundary conditions are useful
when infinite SRR arrays are simulated by HFSS, especially for these symmetrical
SRR unit cells. In this sub-section, we will estimate the resonance frequency of
infinite single square SRR array extending along the electric field direction by using
the equivalent circuit modeling approach. As discussed in section 2.2, the array
effect in magnetic field direction will be ignored as DH is 1.75 mm, so the array in
that direction is sparse. Therefore, the SRR array to be investigated can be
considered approximately as a one dimensional infinite array of single-loop fourgap SRR elements extending along the E field direction (along the y-axis) as
43
shown Figure 3-5. In this figure, each block with impedance Z represents an SRR
unit cell and each block with impedance Zm serves to model the coupling
capacitance between neighboring SRR elements and the ohmic losses occurring due
to leakage currents which flow through the dielectric substrate in the region
between neighboring SRR unit cells [23]. It must be emphasized that the two-port
SRR model shown in Figure 3-2 (c) with equivalent impedance Z (specified in
general by equation (22)) is modified before being used in Figure 3-5 such that the
inductance term Lself is replaced by Lself-2M to account for the effects of subtractive
mutual inductance (M) caused by the neighboring SRR cells on both sides of a
given array member. Couplings from more distant members of the array are
neglected in this model. The mutual inductance effects act in the subtractive manner
as induced currents in all SRR loops of this array are in the same direction as
explained by the magnetic induction phenomena. This situation is simulated by the
mirror imaging effect of the PEC boundary conditions in HFSS computations used
to analyze such infinite SRR arrays. Impedances Z and Zm can be calculated using
equations (53) and (54), respectively. The infinite array can be considered as the
limiting case of a finite length array with n SRR elements as n approaches to
infinity. Then, this finite length array can be represented by a two-port circuit as the
one shown in Figure 2-13, having an equivalent impedance Ztotal in the shunt
branch. As the blocks with impedances Z and Zm can be considered as individual
two-port circuits connected in series, the equivalent impedance of the finite array
can be obtained as in equation (55), where n-1 ≈ n as n approaches to infinity. Then,
the real and imaginary parts of Ztotal can be obtained as in equations (56) and (57).
44
Figure 3-5: Equivalent circuit model for the SRR array which effectively extends to infinity in
the electric field direction.
45
⎛
Rd
Z = ⎜ Rc +
2
⎜
1 + ω 2 C 2 Rd
⎝
⎞
⎟+
⎟
⎠
2
⎛
ω
Rd C
j ⎜ ω ( Lself − 2M ) −
2
⎜
1 + ω 2 C 2 Rd
⎝
⎛ R (1 − jωR C
Z m = ⎜ dm 2 2 dm 2 m
⎜ 1+ ω C R
m
dm
⎝
⎞
⎟
⎟
⎠
(53)
⎞
⎟
⎟
⎠
(54)
Ztotal=n Z +(n-1) Zm
(55)
⎛
Rd
Rdm
Re{Z total } = n⎜ Rc +
+
2
2
2
2
2
2
⎜
1 + ω C Rd
1 + ω C m Rdm
⎝
⎞
⎟
⎟
⎠
(56)
2
2
⎛
Rd C
Rdm C m
⎜
−
Im{Z tota l } = jωn ( Lself − 2M ) −
2
2
⎜
1 + ω 2 C 2 Rd
1 + ω 2 C m Rd m 2
⎝
⎞
⎟
⎟
⎠
(57)
At the resonance frequency, Ztotal becomes real. Then the resonance frequency can
be solved by equating the right hand side of equation (57) to zero. In these
expressions, C is equivalent gap capacitance of an isolated SRR, Lself is the total
self inductance of the metal loop, M is the mutual inductance between adjacent
SRR elements and Cm is the coupling capacitance. Lself and C values are given in
Table 3-2 for the SRR cell used in this chapter. M is calculated using equation (60)
where L is the side length of SRR and d0= 2 DE +w is the centre to centre separation
between the adjacent edges of SRR elements. Rdm is calculated using equation (61)
which is similar to the equation (26) used for calculation of Rd. As discussed earlier
Rd and Rdm are frequency dependent parameters. While solving for the resonance
frequency, Im{Z total } = 0 equation can be written as [23]
aω 4 + bω 2 + d = 0
2
2
(58)
2
a= ( Lself − 2M ) Rd Rdm C m C 2
2
2
2
2
2
2
2
2
b= ( Lself − 2M ) ( Rd C 2 + Rdm C m ) − Rd CRdm C m − Rd C 2 Rdm C m
2
2
d= ( Lself − 2M ) - Rd C − Rdm C m
where
46
(59)
2
d
d
L
L2
M = 0.2 L( In( + 1 + 2 ) − 1 + 02 + 0 ) µH
d0
L
L
d0
(60)
and
Rdm=
2DE
(tan α )(ωε ) Lh
(61)
Table 3-3: Calculated circuit model parameters of the infinite SRR array extending in the
electric field direction
Ccp (49)
17.01*10-15 F
Cpp (51)
0.0407*10-15 F
Cgap =Cpp+Ccp
17.0507*10-15 F
Ctotal =Cgap/4
4.263 * 10-15 F
Lself (52)
58.71* 10-10 H
Cmutualcp (49)
32.625 *10-15 F
Cmutualpp (51)
0.0424*10-15 F
Cmutual
32.67*10-15 F
Cm =2*Cmutual
65.34*10-15 F
M (60)
4.39*10-10 H
f0
35.6 GHz
Parameter values needed in equations (57)-(59) are computed and listed in Table
3-3. Using these values, equation (58) is solved to compute the resonance frequency
to be f 0 =35.6 GHz. As a matter of fact, three more roots are solved from equation
(58), but two of which are turned out to be imaginary and the third one is found to
be negative. Only the fourth root is found meaningful to give f 0 =35.6 GHz. This
47
result is compared to the resonance frequency value found by HFSS simulation of
the infinite SRR array (by using PEC/PMC boundary conditions). As shown in
Figure 3-6, the resonance frequency obtained by the numerical simulation approach
is f 0 =34 GHz. The approximation error is found to be about 4.7 %.
Figure 3-6: Transmission spectrum of the infinite SRR array as estimated by an HFSS
simulation using the PEC/PMC boundary conditions.
3.3
An SRR Array Extending Along the Propagation Direction
In this section an SRR array of size 2x1x1 will be analyzed by both equivalent
circuit modeling approach and HFSS simulations. In the next subsection, the
structure will be investigated as an isolated 2-element array by using PML
boundary conditions in HFSS simulation.
48
3.3.1 Analysis of the Isolated Two-Element Array of Size 2x1x1
In this section, HFSS simulations of the two-element SRR arrays of size 2x1x1 are
performed using PML boundary conditions. Resulting transmission curves are
compared with the results obtained from equivalent circuit modeling approach. In
Figure 3-7, the array structure, its schematic view and the corresponding equivalent
two-port circuit model are shown with substrate dimensions Dx= 8 mm, Dy=4 mm,
h=0.5 mm.
Figure 3-7: a) Geometry of 2x1x1 SRR array, b) Elements of 2x1x1 array, c) Two-port
representation of 2x1x1 array (First alternative for the coupling two-port connection).
In this array topology, mutual inductance between the SRR unit cells is subtractive
due to the direction of the flux linkages. Therefore, the parameter Lself of.Figure 3-2
(c) needs to be modified as Lself-M. The capacitive coupling effect and dielectric
ohmic loss effects occurring in the region between the SRR unit cells can be
modeled as a parallel two-port RC circuit with parameters Cm and Rdm. Two
49
possible configurations can be considered as shown in Figure 3-7 (c) and Figure 3-9
for this coupling two-port circuit. For the first coupling configuration shown in
Figure 3-7 (c), following the similar steps as in section 2.3, we obtain chain
(ABCD) parameters of the circuit.
As this two-element SRR array extends in the direction of propagation, the
equivalent circuit of the overall 2x1x1 array can be considered as the cascade
connection of three equivalent two-port circuits: The SRR two-ports and the
coupling two-port between them. Therefore, the chain parameters of the overall
two-port representation can be computed by making use of equation (13) and
equations (62) through (66) as shown below:
⎡A B⎤
=
⎢C D ⎥
⎣
⎦ cascade
⎡ 1
⎢Y
⎣ firstSRR
0⎤ ⎡ 1
1⎥⎦ ⎢⎣Ymutual
0⎤ ⎡ 1
1⎥⎦ ⎢⎣Ysec ondSRR
0⎤
1⎥⎦
(62)
defining
Ytotal= YfirstSRR +Ymutual+YsecondSRR
(63)
equation (62) becomes
⎡A B⎤
=
⎢C D ⎥
⎣
⎦ cascade
⎡ 1
⎢Y
⎣ total
0⎤
1⎥⎦
(64)
where
YfirstSRR = YsecondSRR =
1
Rc + jω ( Lself − M ) +
Rd
1 + jωRd
C gap
(65)
4
and
Ymutual=
1
Rdm
1 + jωRdm C m
50
(66)
Finally, using the conversion rules between chain and S- parameters [21], the S21
parameter can be computed from equation (67) given below
S 21 =
2
A+
B
+ CZ 0 + D
Z0
(67)
As stated previously in section 2.3, frequency dependent Z0 array data are taken
from the HFSS simulation output which is executed with PML boundary
conditions. The values of Cgap/4, Lself, M and Cm are taken from Table 3-3. The
parameters Rdm, Rd and Rc are all frequency dependent and their values are
calculated by using equations (61), (26) and (25), respectively. In Figure 3-8, the
HFSS result along with the result of equivalent circuit model approach (computed
by MATLAB codes written in this thesis) are used to plot transmission spectra of
the 2x1x1 SRR array. The transmission spectrum is given by the S 21 curve and the
value of transmission can not exceed unity (corresponding to the zero decibel
level). Positive logarithmic transmission values seen in Figure 3-8 are not
physically meaningful but they are spurious effects due to the use of PML boundary
conditions. Also, the transmission spectrum obtained by HFSS simulation indicates
much higher losses which is an expected result as PMLs are highly lossy structures.
Two transmission curves shown in Figure 3-8 are similar around the resonance
frequency which is very close to 34 GHz.
51
Figure 3-8: Estimated transmission spectrum of 2x1x1 SRR array using HFSS simulations
with PML boundary conditions and DH=1 mm and using the equivalent two-port circuit given
in Figure 3-7(c).
The second possible coupling configuration is shown in Figure 3-9 where the
coupling two-port circuit has a parallel CmRdm combination in its series branch.
Now, the ABCD parameters of the overall two-port network can be obtained as
given in equations (68) and (69) below:
⎡A B⎤
=
⎢C D ⎥
⎣
⎦ cascade
⎡A
⎢C
⎣
⎡ 1
⎢Y
⎣ firstSRR
0 ⎤ ⎡1
⎢
1⎥⎦ ⎢0
⎣
1
⎤
1
Ymutual ⎥ ⎡⎢
1 ⎥⎦ ⎣Ysec ondSRR
0⎤
1⎥⎦
Y
1
⎡
⎤
1 + sec ondSRR
⎢
B⎤
Ymutual
Ymutual ⎥
⎢
⎥
=
Y firstSRRYsec ondSRR
Y firstSRR ⎥
D ⎥⎦ cascade ⎢Y
+ Ysec ondSRR 1 +
⎢ firstSRR +
Ymutual
Ymutual ⎥⎦
⎣
52
(68)
(69)
After converting these chain parameters to S-parameters [21], the estimated
transmission spectrum of the 2x1x1 array is obtained and plotted in Figure 3-10.
The transmission spectrum obtained by HFSS with PML boundary conditions is
also plotted in the same figure for comparison. The spurious effects caused by PML
usage are observed in these results. However, both transmission curves indicate the
expected resonance around 34 GHz.
Figure 3-9: Two-port representation of the 2x1x1 array (Second alternative for the coupling
two-port connection).
53
Figure 3-10: Estimated transmission spectrum of the 2x1x1 SRR array using HFSS
simulations with PML boundary conditions and DH=1 mm. and using the equivalent two-port
circuit given in Figure 3-9.
3.3.2 Numerical Simulation and Equivalent Circuit Modeling of Two
Layer Infinite SRR Array
The SRR array to be examined in this subsection is assumed to be infinitely large in
the E field and H field directions. It has only two layers along the propagation
direction. The array parameters DE and DH are taken as 0.6 mm and 1.75 mm,
respectively, to have an array which is sparse along the H field direction.
Therefore, the array can be considered as an mxnxp array, effectively, where m=2,
p=1 (approximately) and n approaches to infinity. Such an array can be simulated
by HFSS using PEC/PMC boundary conditions as discussed in section 3.2.2. An
approximate equivalent two-port circuit model for this array is shown in Figure
3-11 where the capacitive coupling and dielectric losses in the region between
54
neighboring SRR cells along the propagation direction are neglected for simplicity.
The inductive coupling between such SRR cells, however, is taken into account by
the (Lself-3M-2Mc) term. Here M is the mutual inductance between two adjacent
SRR cells along the incident electric field and propagation directions. Mc is the
mutual inductance between two SRR cells in cross locations. The parameter Mc is
computed using equation (60) with d0= 2 (2 D E + w) . Mathematical expressions
for the impedances Z and Zm are given in equations (70) and (71), respectively.
Following the approach used in section 3.2.2., the real and imaginary parts of the
equivalent impedance Ztotal for the overall two-port array representation are
obtained as in equations (72) and (73). The resonance frequency of the array can be
obtained by solving equations (74) and (75) after equating the imaginary part of
Ztotal to zero.
55
I1
I1
+
+
L self - 3M - 2M c
L self - 3M - 2M c
Rc
Z
Rc
Rd
Cgap / 4
Cm
Zm
R dm
Cm
Rc
Cm
Zm
Rc
Rd
Cgap / 4
goes to infinity
R dm
L self - 3M - 2M c
L self - 3M - 2M c
Z
Rd
Cgap / 4
Cgap / 4
R dm
Cm
Rd
R dm
goes to infinity
Figure 3-11: An equivalent circuit model for the double layer SRR array (m x n x p) where
m=2, p=1 (due to sparse array approximation) and n approaches to infinity.
56
Rd
1 ⎡⎛
Z = ⎢⎜ Rc +
2
2 ⎢⎣⎜⎝
1 + ω 2 C 2 Rd
⎞
⎟+
⎟
⎠
⎛
ωRd 2 C
j ⎜ ω ( Lself − 3M − 2M c ) −
2
⎜
1 + ω 2 C 2 Rd
⎝
⎞⎤
⎟⎥
⎟
⎠⎥⎦
(70)
⎛ R (1 − jωRdm C m ⎞
⎟
Z m = ⎜ dm
⎜ 2(1 + ω 2 C 2 R 2 ) ⎟
m
dm
⎝
⎠
(71)
⎛ Rc
⎞
Rd
Rdm
⎟
+
+
Ztotal (real)= n⎜⎜
2
2
2 ⎟
2
2
2
2
2(1 + ω C Rd ) 2(1 + ω C m Rdm ) ⎠
⎝
(72)
2
2
⎛ L self − 3M − 2M c
⎞
Rd C
R dm C m
⎜
⎟
−
−
Ztotal(im)= jωn⎜
2
2
2 ⎟
2
2
2
2
ω
C
R
ω
C
R
+
+
2
(
1
)
2
(
1
)
d
m
dm
⎝
⎠
(73)
aω 4 + bω 2 + d = 0
(74)
2
2
a=(Lself-3M-2Mc) Rd 2 Rdm C m C 2
2
2
2
2
2
2
2
2
b=(Lself-3M-2Mc) ( R d C 2 + R dm C m ) − R d CR dm C m − R d C 2 R dm C m
2
(75)
2
d=(Lself-3M-2Mc)- R d C − R dm C m
The values of parameters in equation (75) are already calculated in section 3.2.2
and given in Table 3-3. Only the parameter Mc is used for the first time here and it
is calculated using equation (60).All of the resulting values are given in Table 3-4.
The only meaningful root of equation (74) is found to be 39.7 GHz with about 12.5
percent error as compared to the resonance frequencies (34 GHz and 35.3 GHz)
estimated by the HFSS simulation with PEC/PMC boundary conditions. The error
in the equivalent circuit model estimation for the resonance frequency can be
explained by the fact that coupling effects along the H field direction and more
importantly, the capacitive coupling and loss effects along the propagation direction
are neglected in the equivalent circuit model. It should be also indicated that the
present approximate two-port model can estimate only one resonance frequency,
misses the other. The transmission spectrum of the SRR array is shown in Figure
3-12.
57
Table 3-4: Calculated model parameters for the infinite SRR array in electric field direction
Cgap/4
4.263 * 10-15 F
Lself (52)
58.71* 10-10 H
Cmutual
32.67*10-15 F
Cm (2*Cmutual)
65.34*10-15 F
M (60)
4.39*10-10 H
Mc(60)
3.3*10-10 H
f0
39.7 GHz
Figure 3-12: Simulation of 2x1x1 array with PEC/PMC boundary.
3.4
An SRR Array Extending Along the Electric Field Direction
In this section, an SRR array of size 1x2x1 will be analyzed over the frequency
band from 25 GHz to 40 GHz. First, this small array will be considered in isolation
58
and its transmission spectrum will be obtained by HFSS using PML boundary
conditions. Then, it will be analyzed by using PEC/PMC boundary conditions to
simulate an infinite SRR array.
3.4.1 Analysis of the Isolated Two-Element Array of Size 1x2x1
The 1x2x1 SRR array and its equivalent two-port model are shown in Figure 3-13
where the substrate dimensions are Dx= 4 mm, Dy=8 mm, h=0.5 mm.
59
Figure 3-13: a) Geometry of the 1x2x1 array, b) Elements of 1x2x1 array, c) Equivalent twoport circuit model for the 1x2x1 array.
60
All of the circuit parameters shown in Figure 3-13 (c) are calculated in section 3.2.2
and given in Table 3-3. In addition, expressions of loss parameters Rc, Rd and Rdm,
are given in equations (25),(26) and (61), respectively. To obtain the transmission
spectrum of the 1x2x1 array, ABCD parameters of the equivalent two-port circuit
need to be obtained. Since the two-port circuit models representing individual SRRs
and mutual effects are connected in series along the electric field direction, their
individual Z parameter matrices are added to get the overall Z matrix of the array.
Chain matrices of each SRR unit cell are the same and given by equation (76) where
the admittance term Y is computed using equation (78). The chain matrix of the
coupling circuit is determined by the help of equations (77) and (79). Then, the
overall Z-matrix of the 1x2x1 array two-port is obtained using equations (80)
through (84).
⎡A B⎤
⎡A B⎤
⎡ 1 0⎤
=⎢
=⎢
⎢C D ⎥
⎥
⎥
⎣
⎦ firstSRR ⎣C D ⎦ sec ondSRR ⎣Y 1⎦
(76)
⎡A B⎤
⎡ 1 0⎤
=⎢
⎢C D ⎥
⎥
⎣
⎦ mutual ⎣Ym 1⎦
(77)
where
1
Y=
Rd
Rc + jω ( Lself − M ) +
1 + jωRd
C gap
(78)
4
and
Ym=
1
Rdm
1 + jωRdm C m
Then, using the general conversion rule:
61
(79)
⎡ Z 11
⎢Z
⎣ 21
AD − BC ⎤
⎥
C
⎥
D
⎥
C
⎦
⎡A
Z 12 ⎤ ⎢ C
=⎢
Z 22 ⎥⎦ ⎢ 1
⎣C
(80)
Z matrices of the SRR cells and the coupling two-port are given as:
⎡ Z 11
⎢Z
⎣ 21
Z 12 ⎤
=
Z 22 ⎥⎦ firstSRR
⎡ Z 11
⎢Z
⎣ 21
⎡1
Z 12 ⎤
⎢
= ⎢Y
⎥
Z 22 ⎦ sec ondSRR ⎢ 1
⎣Y
1⎤
Y⎥
1⎥
⎥
Y⎦
(81)
and
⎡ Z 11
⎢Z
⎣ 21
Z 12 ⎤
Z 22 ⎥⎦ mutual
⎡1
⎢Y
=⎢ m
⎢1
⎢⎣ Ym
1⎤
Ym ⎥
⎥
1⎥
Ym ⎥⎦
(82)
Then, defining:
Ztotal=ZfirstSRR+Zmutual+ZsecondSRR
(83)
Z matrix of the overall array two-port is found as:
[Z total ] =
⎡2 1
⎢Y + Y
m
⎢
2
1
⎢ +
⎢⎣ Y Ym
2 1⎤
+
Y Ym ⎥
⎥
2 1⎥
+
Y Ym ⎥⎦
(84)
Next, the complex S-parameters are obtained using the [Z total ] matrix [21]. The
transmission curve ( S 21 versus frequency curve) obtained via equivalent circuit
modeling is shown in Figure 3-14. This array structure is also analyzed with HFSS
using PML boundary conditions with D H =1.75 mm and the resulting curve is also
plotted in Figure 3-14 for comparison. This curve looks like a compressed and lossy
version of the transmission curve obtained by equivalent modeling. Resonance
62
frequencies are found to be 33.6 and 34.6 GHz from the circuit model approach
whereas the resonances are observed at 34.6 and 35.1 GHz according to the HFSS
simulation results.
Figure 3-14: Estimated transmission spectra of the 1x2x1 SRR array using HFSS simulations
with PML boundary conditions and D H =1.75 mm and using the equivalent two-port circuit
given in Figure 3-13 (c).
3.4.2 Numerical Simulation and Equivalent Circuit Modeling of Infinite
SRR Array Along the Electric Field Direction
When PEC/PMC boundary conditions are used instead of PML boundary condition,
an infinite array is formed in both E field and H field directions. As DH is chosen
to be a relatively large value, the array is sparse in H field direction. Hence, the
63
resulting array can be considered as an infinitely long, one-dimensional array of
size 1x ∞ x1. In this section, the resonance frequency of this infinite SRR array will
be calculated and compared with the resonance frequency obtained from the HFSS
simulation results. In section 3.2.2, the resonance frequency of this infinite SRR
array has already been calculated to be 35.6 GHz. The transmission spectrum of this
array obtained by using HFSS with PEC/PMC boundary conditions is shown in
Figure 3-15, which is the exact replica of Figure 3-6. The resonance frequency
obtained by HFSS simulations is 34 GHz and very close to the result of 35.6 GHz,
which is obtained from the equivalent circuit approach.
Figure 3-15: Transmission spectrum of the 1x2x1 SRR array obtained by HFSS simulation
with PEC/PMC boundary conditions.
3.5
The 2x2x1 SRR Array
Combining two arrays of sizes 2x1x1 and 1x2x1, it is possible to form a 2x2x1
array. The geometry, dimensions and some parameters of this array are shown in
Figure 3-16, where the substrate dimensions are Dx=Dy=8 mm and h=0.5 mm.
64
Geometry, dimensions and material properties of the SRR unit cell used in this
array are the same as those used earlier in Chapter 3. In this section, we will first
examine this 2x2x1 array in isolation. Then, we will extend investigation to an
infinite array form.
65
Figure 3-16: a) Unit cell geometry of 2x2x1 array, b) Parameters of the 2x2x1 SRR array.
66
3.5.1 Analysis of the Isolated Four Element Array of Size 2x2x1
Equivalent two-port circuit model of the 2x2x1 array is shown in Figure 3-17. In
this model, cross mutual inductance effects are also considered and included in the
model via the parameter Mc. The SRR equivalent circuit shown in Figure 3-2 (c),
the coupling circuit (along the E field direction) shown in Figure 3-5 and the
coupling circuit alternative (along the propagation direction) shown in Figure 3-7
(c) are used to form the equivalent two-port circuit model of 2x2x1 array shown in
Figure 3-17. The chain parameter matrix and the corresponding Z-parameter matrix
of the overall 2x2x1 array structure are computed using equations (85) through (93).
Then, the complex S parameter (S21) is computed from the Z-parameters using
equation (67). The S 21 versus frequency curve of the array is plotted in Figure
3-18. The transmission spectrum obtained using HFSS with PML boundary
conditions ( DH =1.75 mm) is also plotted in the same figure for comparison.
67
Figure 3-17: Two-port model for 2x2x1 SRR array.
68
⎡ 1
⎡A B⎤
⎡A B⎤
=⎢
=⎢
⎢C D ⎥
⎥
⎣
⎦ upperSRRblock ⎣C D ⎦ lowerSRRblock ⎣2Y + Ym
⎡ 1
⎡A B⎤
=⎢
⎢C D ⎥
⎣
⎦ mutual ⎣2Ym
0⎤
1⎥⎦
0⎤
1⎥⎦
(85)
(86)
where
1
Y=
Rd
Rc + jω ( Lself − 2 M − M c ) +
1 + jωR d
C gap
(87)
4
and
Ym=
1
Rdm
1 + jωRdm C m
(88)
Using
⎡ Z 11
⎢Z
⎣ 21
⎡A
Z 12 ⎤ ⎢ C
=⎢
Z 22 ⎥⎦ ⎢ 1
⎣C
AD − BC ⎤
⎥
C
⎥
D
⎥
C
⎦
(89)
Z parameter matrices for the SRR unit cell and coupling circuit two-ports can be
obtained as:
⎡ Z 11
⎢Z
⎣ 21
Z 12 ⎤
=
Z 22 ⎥⎦ firstSRR
⎡ Z 11
⎢Z
⎣ 21
⎡
⎢ 2Y
Z 12 ⎤
⎢
=
Z 22 ⎥⎦ sec ondSRR ⎢
⎢⎣ 2Y
and
69
1
+ Ym
1
+ Ym
1 ⎤
2Y + Ym ⎥
⎥
1 ⎥
2Y + Ym ⎥⎦
(90)
⎡ 1
⎢ 2Y
Z 12 ⎤
=⎢ m
⎥
Z 22 ⎦ mutual ⎢ 1
⎢⎣ 2Ym
1 ⎤
2Ym ⎥
⎥
1 ⎥
2Ym ⎥⎦
(91)
Ztotal=Zfirstblock+Zmutual+Zsecondblock
(92)
⎡ Z 11
⎢Z
⎣ 21
Defining
The equivalent Z parameter matrix of the overall array can be obtained as:
[Z total ] =
⎡
⎢ 2Y
⎢
⎢
⎢⎣ 2Y
2
1
+
+ Ym 2Ym
2
1
+
+ Ym 2Ym
2
1 ⎤
+
2Y + Ym 2Ym ⎥
⎥
2
1 ⎥
+
2Y + Ym 2Ym ⎥⎦
(93)
Figure 3-18: Estimated transmission spectra of the 2x2x1 SRR array using HFSS simulations
with PML boundary conditions and D H =1.75 mm and using the equivalent circuit model
shown in Figure 3-17.
70
3.5.2 Numerical Simulation and Equivalent Circuit Modeling of Infinite
Two Layer SRR Array
When the 2x2x1 array is simulated by HFSS using PEC/PMC boundary conditions,
the resulting infinite array with two layers along the propagation direction turns out
to be exactly the same array examined in section 3.3.2. Therefore, exactly the same
transmission curve shown in Figure 3-19 and the same resonance frequencies at 34
GHz and 35.3 GHz are obtained with HFSS simulations. Also a resonance
frequency of 39.7 GHz is estimated by the equivalent circuit model. As the
capacitive coupling between the SRR cells in the propagation direction is neglected,
only one resonance frequency is estimated with this approach.
Figure 3-19: Transmission spectrum for the 2x2x1 SRR array estimated by HFSS with
PEC/PMC boundary conditions.
71
CHAPTER 4
NUMERICAL SIMULATIONS, MEASUREMENTS
AND TWO PORT EQUIVALENT CIRCUIT
MODELING OF SQUARE SHAPED SINGLE LOOP
SRR STRUCTURES
4.1
Introduction
In this chapter, a single square loop SRR with four splits and its 4x1x1 array are
investigated. In addition to HFSS simulations and equivalent circuit modeling
results, these structures are also fabricated and their transmission spectra are
measured. Due to the limitations in our measurement setup, the SRR unit cell is
designed to resonate around 12 GHz and the measurements are carried on over the
[10 GHz -13] GHz frequency band.
Transmission spectrum of the SRR unit cell is obtained by HFSS using both PML
and PEC type boundary conditions as shown in Figure 4-1. The isolated unit cell
structure is analyzed when PML boundaries are used. Behavior of a single layer (in
the propagation direction) infinite SRR array within a metallic waveguide is
simulated when PEC boundary conditions are employed. The same approach is also
used to investigate the behavior of the 4x1x1 SRR array as to be presented in this
chapter. As the transmission spectra of these structures are measured within a
metallic waveguide, PEC boundary conditions are used in the associated HFSS
simulations for meaningful comparison of the results.
72
In the fabrication of the SRR unit cell and the 4x1x1 array, copper inclusions are
formed over the low-loss dielectric substrate Arlon AD350, which has a thickness
of h=0.762 mm and relative permittivity of єr =3.5. Thickness of copper strips is
t=0.035 mm.
Figure 4-1: Boundary conditions used in HFSS simulations: a) Use of PML conditions to
examine the isolated SRR cell, b) Use of PEC boundary conditions to simulate SRR behavior
within a metallic waveguide.
4.2
Analysis and Measurements of the SRR Unit Cell
Geometrical parameters of the fabricated SRR unit cell are shown in Figure 4-2.
Values of these geometrical parameters are given in Table 4-1.
73
Figure 4-2: SRR unit cell a) Geometry and excitation, b) Design parameters.
Table 4-1: Geometrical parameters of the fabricated SRR unit cell
Substrate
L (Side length of
Dimensions
the SRR)
Dx=Dy=10mm
h=0.762 mm
g (Gap Width)
8 mm
0.3 mm
w (SRR
Width)
0.6 mm
4.2.1 Analysis of the Isolated SRR Unit Cell
Two-port equivalent circuit model for the isolated SRR unit cell is shown in Figure
4-3. The values of calculated circuit elements are given in Table 4-2. The
equivalent loss resistance values for Rc and Rd are calculated using equations (25)
74
and (26) as functions of frequency. The transmission spectrum of the unit cell is
computed using equation (24) based on this equivalent circuit model. Also, the
transmission spectrum of this SRR unit cell is computed using HFSS with PML
boundary conditions. Transmission curves obtained using both approaches are
plotted in Figure 4-4. In this figure, it is observed that the resonance frequency
values of these two transmission curves are very close to each other, both of which
are approximately at 11.90 GHz. The overall shapes of both transmission curves are
sufficiently close to each other as well.
Figure 4-3: Equivalent two-port circuit representation of the SRR unit cell.
Table 4-2: Model parameters of the fabricated SRR unit cell
Ccp (49)
35.483*10-15 F
Cpp (51)
2.17*10-15 F
Cgap (Cpp+Ccp)
37.652*10-15 F
Ctotal (Cgap/4)
9.413 * 10-15 F
Lself (52)
191* 10-10 H
75
Figure 4-4: Estimated transmission spectra of the SRR unit cell using HFSS simulations with
PML boundary conditions and DH=5 mm. and using the equivalent two-port circuit given in
Figure 4-3.
4.2.2 Numerical
Simulation,
Equivalent
Circuit
Modeling
and
Measurement of the SRR Unit Cell within Waveguide
As indicated earlier, transmission curve of the fabricated SRR structures are
measured within a waveguide. Therefore, to account for the imaging effect of the
metallic walls of the waveguide, PEC type boundary conditions are used in HFSS
simulations while estimating the transmission spectrum of the measured structures.
In other words, a single layer (in propagation direction) infinite SRR array is
effectively created by these simulations and measurements when the SRR unit cell
is placed within a waveguide. As the separation distance between the SRR surface
and PEC walls of the waveguide is large along the H field direction, the resulting
76
array can roughly be considered as a one-dimensional infinite SRR array extending
in the E field direction. Equivalent circuit model of the resulting array structure is
suggested as shown in Figure 4-5 based on this one-dimensional periodical array
assumption. The equivalent circuit model in this figure has been previously
suggested and studied in section 3.2.2.
For the SRR parameters and materials used in the fabrication of the SRR unit cell,
parameters of the two-port equivalent circuit are calculated and listed in Table 4-3.
Then using the equations (53) through (61), resonance frequency of the effective
SRR array structure is estimated to be 13.3 GHz as shown in Table 4-3. The
transmission curve obtained with HFSS using PEC boundary conditions, on the
other hand, is given in Figure 4-6 which shows a resonance at 11.95 GHz. Finally,
the measured transmission spectrum is plotted in Figure 4-7 indicating a resonance
frequency of 12 GHz.
In conclusion, the measurement and HFSS simulation results are found in very
good agreement. The error made in the estimation of the resonance frequency by
the equivalent circuit model approach is about ten percent and it can be explained
by the fact that coupling effects along the H field direction are neglected in this
model.
77
Figure 4-5: Equivalent circuit model of the effective SRR array formed by placing SRR unit
cell within the waveguide in measurements.
78
Table 4-3: Calculated model parameters of the infinite SRR array extending along the incident
electric field direction
Ccp (49)
35.483*10-15 F
Cpp (51)
2.17*10-15 F
Cgap (Cpp+Ccp)
37.652*10-15 F
Ctotal (Cgap/4)
9.413 * 10-15 F
Lself (52)
191* 10-10 H
Cmutualcp (49)
84.76 *10-15 F
Cmutualpp (51)
2.085*10-15 F
Cmutual
86.85*10-15 F
Cm (2*Cmutual)
173.7*10-15 F
M (60)
17.9*10-10 H
f0
13.3 GHz
Figure 4-6: Simulated transmission spectrum of the fabricated SRR unit cell using HFSS with
PEC boundary conditions.
79
Figure 4-7: Measured transmission spectrum of the fabricated SRR unit cell when it is placed
within a measurement waveguide.
4.3
Analysis and Measurements of the 4x1x1 SRR Array
The basic geometry and parameters of the four-element SRR array of size 4x1x1 is
shown in Figure 4-8 where the capacitive coupling effects are also schematically
indicated in Figure 4-8(b). This small sized SRR array is also investigated with
HFSS simulations, equivalent circuit model and measurements both in isolation and
within a waveguide in subsections 4.3.1 and 4.3.2, respectively.
80
Figure 4-8: Four element SRR array of size 4x1x1 a) Geometry and excitation, b) Capacitive
coupling effects along the propagation direction.
4.3.1 Analysis of the Isolated 4x1x1 SRR Array
The equivalent circuit model suggested for the isolated 4x1x1 array is seen in
Figure 4-9. In this two-port model, all of the mutual inductance and capacitance
effects along with losses are considered using the coupling two-port model
suggested earlier in section 3.3.1 and Figure 3-7 (c). The only difference between
the equivalent circuit models given in Figure 3-7 (c) and Figure 4-9 is that the
inductance term L-M is replaced by L-2M for the SRR unit cells which have
neighboring SRR cells on their both sides.
Following the same computational steps as in section 3.3.1 and using equations (94)
through (98) (which are the adapted versions of the equations of (62) through (66) in
this four-element array case), the transmission spectrum curve is computed (using
equation (67) for S21 calculation also ) and plotted in Figure 4-10. Transmission
spectrum curve of the same isolated SRR array is also obtained by HFSS using
81
PML boundary conditions and plotted in Figure 4-10 for comparison. Ignoring the
spurious effects stemming from the use of PML, both curves in this figure resemble
each
other
around
the
expected
resonance
82
frequency
of
11.90
GHz.
Figure 4-9: Two-port equivalent circuit model suggested for the isolated 4x1x1 SRR array
Figure 4-9: Two-port equivalent circuit model suggested for the isolated 4x1x1 SRR array.
83
⎡1
⎡A B⎤
=⎢
⎢C D ⎥
⎣
⎦ cascade ⎣Y1
0⎤ ⎡ 1
1⎥⎦ ⎢⎣Ym
0⎤ ⎡ 1
1⎥⎦ ⎢⎣Y2
⎡A B⎤
=
⎢C D ⎥
⎣
⎦ cascade
0⎤ ⎡ 1
1⎥⎦ ⎢⎣Ym
0⎤ ⎡ 1
1⎥⎦ ⎢⎣Y3
0⎤ ⎡ 1
1⎥⎦ ⎢⎣Ym
1
⎡
⎢Y + Y + Y + Y + 3Y
2
3
4
m
⎣ 1
0⎤
1⎥⎦
0⎤ ⎡ 1
1⎥⎦ ⎢⎣Y4
0⎤
1⎥⎦
(94)
(95)
where
Y1=Y4 =
1
Rd
Rc + jω ( Lself − M ) +
1 + jωRd
Y2=Y3 =
C gap
(96)
4
1
Rc + jω ( Lself − 2M ) +
Rd
1 + jωRd
C gap
(97)
4
and
Ym=
1
Rdm
1 + jωRdm C m
84
(98)
Figure 4-10: Estimated transmission spectrum of the isolated 4x1x1 SRR array using HFSS
with PML boundary conditions and using the equivalent two-port model given in Figure 4-9.
4.3.2 Measurements and Numerical Simulations for the 4x1x1 SRR
Array Placed Within a Waveguide
In this section, transmission spectrum of the 4x1x1 SRR array is both measured and
estimated using HFSS (with PEC boundary conditions) within the measurement
waveguide. As discussed in section 4.3.1, the effect of placing this SRR array in a
waveguide is to create a four-layer (in the propagation direction) infinite SRR array
both in E field and H field directions. The transmission curves obtained by HFSS
simulations and by laboratory measurements are given in Figure 4-11 and Figure
4-12, respectively. Except for a shift in frequency, the overall variations of these
two curves are very similar to each other with resonance frequencies observed at
11.72 and 11.96 GHz in HFSS simulation results and 12 and 12.4 GHz in the
measurement results.
85
Pictures of the fabricated SRR unit cell and the 4x1x1 SRR array are shown in
Figure 4-13 and Figure 4-14. Picture of a simple open-air measurement set-up using
a pair of probes and a network analyzer is shown in Figure 4-15. It should be noted
that this is not the actual waveguide-measurement set-up used in our experimental
studies. This picture is given just for giving a rough idea about the open air
measurement set-up.
Figure 4-11: Transmission spectrum of the 4x1x1 array simulated by HFSS with PEC
boundary conditions.
86
Figure 4-12: Measured (within a waveguide) transmission spectrum of the 4x1x1 SRR array.
Figure 4-13: Fabricated SRR unit cell and probes used to measure its transmission spectrum.
87
Figure 4-14: Comparison of the fabricated SRR unit cell and 4x1x1 SRR array.
Figure 4-15: A simple experimental set-up used to measure transmission spectrum of SRR
structures in air.
88
CHAPTER 5
CONCLUSIONS AND FUTURE WORK
Split ring resonator (SRR) is one of the main sub-wavelength components used to
design left handed metamaterials in microwave or optical frequencies. SRR arrays
provide engineered materials with negative values of effective permeability. In this
thesis, a simple and fully symmetrical SRR unit cell topology composed of a single
square-shaped metallic loop with four splits at the middle of each edge is
investigated. Geometrical simplicity is preferred in the design of such subwavelength structures especially in millimeter wave, terahertz and upper optical
frequency ranges. Therefore, investigations in this thesis are restricted to this simple
but practically useful SRR unit cell geometry.
The basic purpose of this thesis research has been to establish equivalent lumped
element circuit models to describe the two-port circuit characterization of SRR
arrays with special emphasis on cell-to-cell coupling effects. Obviously, such
equivalent circuit representations can describe actual array behavior only
approximately. On the other hand, full-wave electromagnetic solvers making use of
numerical methods such as the FEM (finite elements method) or FDTD (finite
difference time domain) method provide much better accuracy in the expense of
highly increased computer memory and run time requirements especially for larger
array sizes. Availability of sufficiently accurate equivalent circuit models for SRR
arrays offer an approximate yet computationally efficient and fast way of analysis
for finite but large size arrays. This approach would also make the solution of
optimization problems possible for the design of metamaterials with specific design
requirements.
89
In this thesis, the basic SRR unit cell topology is represented by a two-port
equivalent circuit model having a series RLC resonant circuit in its shunt branch.
Other than the self inductance of the metallic loop and the total capacitance
accounting for the equivalent effect of four individual gap capacitances (all
connected in series), the model has two equivalent resistances; a resistor serially
connected to the inductor to model conductor losses and another resistor connected
in parallel to the capacitor to model ohmic losses occurring within the dielectric
substrate. Values of the self inductance of the metallic loop, equivalent capacitance
and the values of loss resistances are computed using the geometrical and material
properties of the SRR unit cell. The electromagnetic coupling effects between two
neighboring cells of a given SRR array in the electric field and in the propagation
directions are modeled by a proper RC two-port coupling circuit in addition to the
mutual inductance parameter which is used to modify the self inductance values of
individual cells.
In this thesis, the basic SRR equivalent circuit model and the coupling circuit twoport models are used to estimate the transmission spectra and resonance frequencies
of various array forms over two different frequency bandwidths; 10 GHz to13 GHz
band and 25 GHz to 40 GHz band. The finite size small arrays such as 1x2x1,
2x1x1, 2x2x1 and 4x1x1 arrays are simulated in isolation using HFSS with PML
boundary conditions. Transmission spectrum results from such simulations are
compared to equivalent circuit model results with good agreement especially
around the resonance frequency of the simulation bandwidth. Excess losses and
unexpected spurious effects are observed due to the use of PML boundary
conditions over very wide bandwidths. Results of these HFSS simulations are also
found to be very sensitive to the location of PML boundaries. Alternatively, such
basic arrays are turned into infinite array structures by using PEC/PMC boundary
conditions in HFSS simulations. This situation is also experimentally implemented
by measuring an SRR unit cell and its 4x1x1 SRR array within a waveguide due to
the mirror imaging effect of metallic walls of the guide. Resulting infinite array
effects are taken into account in equivalent circuit models as well. In the
90
investigation of these array structures, the periodicity along the applied magnetic
field direction is taken to be large enough to omit the array effects coming from that
dimension. Then, the resulting arrays are simulated to be of infinite size along the
applied electric field direction with one or more layers along the propagation
direction of the incident excitation.
In a given array problem, the overall SRR array is eventually represented by a
single equivalent two-port circuit by considering the series, parallel or cascade
connections of smaller two-port units belonging to SRR unit cells or coupling
circuits. The complex S-parameter matrix of the overall array two-port is obtained
via conversions from Z-matrix or chain matrix representations appropriately. All
these computations are carried on by simple and fast MATLAB codes written
during this thesis research. The transmission spectrum (i.e. the S 21 ( f ) versus
frequency (f) curve) estimated by this approach is found to be in very good
agreement with transmission spectra obtained both by HFSS simulations and
measurements.
For the purpose of experimental verification, two of the investigated structures, the
SRR unit cell resonating around 12 GHz in the X-band and a 4x1x1 SRR array
(formed by four of such cells in the propagation direction) are fabricated by copper
stripes printed over a low loss dielectric substrate. Complex scattering parameters
of these two structures are measured within a metallic waveguide. Resulting
transmission spectra are compared with the transmission curves obtained by HFSS
simulations where PEC boundary conditions are applied at the surfaces of the
computational volume (surfaces which coincide with the walls of the waveguide
used in the measurement setup). Again, simulation results, equivalent circuit
estimations and measurement results are found in good agreement.
As future work, the novel equivalent two-port circuit models suggested in this
thesis can be further improved especially by including the coupling effects along
the magnetic field direction as well. Effect of asymmetry in the SRR unit cell
91
geometry, concerning the number or locations of gaps for instance, may be
investigated in circuit modeling. Equivalent circuit models of SRR structures under
different excitation conditions can be studied. Finally, the suggested equivalent
circuit modeling approach can be applied to SRR array optimization to realize
special design requirements.
92
REFERENCES
1.
V.G. Veselago, “The Electrodynamics of Substances with Simultaneously
Negative Values of Permittivity and Permeability,” Sov. Phys. Uspekhi,
Vol.10, 1968.
2.
J. B. Pendry, A. J. Holden, D. J. Robbins, and W. J. Stewart, “Magnetism
from Conductors and Enhanced Nonlinear Phenomena,” IEEE Trans.
Microwave Theory Tech., Vol. 47, No. 11, 1999.
3.
D. R. Smith, W. J. Padilla, D. C. Vier, S. C. Nemat-Nasser, and S. Schultz,
“Composite Medium with Simultaneously Negative Permeability and
Permittivity,” Physical Review Letters, Volume 84, Number 18,1 May 2000.
4.
R. Marqués, F. Medina, and R. Rafii-El-Idrissi, “Role of Bianisotropy in
Negative Permeability and Left-Handed Metamaterials,” Phys. Rev. B, Vol.
65, 144440, 2002.
5.
R. Marqués, F. Mesa, J. Martel, and F. Medina, “Comparative Analysis of
Edge- and Broadside-Coupled Split Ring Resonators for Metamaterial Design
Theory and Experiments,” IEEE Transactions on Antennas and Propagation,
Vol. 51, No. 10, Oct. 2003.
6.
F. Bilotti, A. Toscano, L. Vegni, K. Aydin, K. B. Alici, and E. Ozbay,
“Equivalent-Circuit Models for the Design of Metamaterials Based on
Artificial Magnetic Inclusions,” IEEE Transactions on Microwave Theory and
Techniques, Vol. 55, No. 12, December 2007.
7.
I.I. Smolyaninov, Y.J. Hung, C.C. Davis, “Magnifying Superlens in the
Visible Frequency Range,” Science, 315, 1699-1701, 2007.
93
8.
X. Zhang, Z. Liu, “Superlenses to Overcome the Diffraction Limit,” Nature
Mater, 7, 435-441, 2008.
9.
J. B. Pendry, D. Schurig D.R. Smith, “Controlling Electromagnetic Fields,”
Science 312, 1780-1782, 2006.
10. J. Baena, J. Bonache, F. Martín, R. M. Sillero, F. Falcone, Tx. Lopetegi, M.
A. G. Laso, J. García–García, I. Gil, M. Flores Portillo and M. Sorolla,
“Equivalent-Circuit Models for Split-Ring Resonators and Complementary
Split-Ring Resonators Coupled to Planar Transmission Lines,” IEEE
Transactions on Microwave Theory and Techniques, Vol. 53, No. 4, April
2005.
11. Q. Wu, M. Wu and F. Meng, “SRRs’ Artificial Magnetic Metamaterials
Modeling Using Transmission Line Theory,” Electromagnetics Research
Symposium 2005, Hangzhou, China, August 22-26.
12. N.P. Johnson, A.Z. Khokhar, H.M.H. Chong, R.M. De La Rue and S.
McMeekin, “Characterisation at Infrared Wavelengths of Metamaterials
Formed by Thin-Film Metallic Split-Ring Resonator Arrays On Silicon,”
Electronics Letters 14th September 2006 Vol. 42 No. 19.
13. J. Wang, S. Qu, Z. Xu, H. Ma, Y. Yang, and C. Gu, “A Controllable Magnetic
Metamaterial: Split-Ring Resonator With Rotated Inner Ring,” IEEE
Transactions on Antennas and Propagation, Vol. 56, No. 7, July 2008.
14. O. Sydoruk, E. Tatartschuk, E. Shamonina, and L. Solymar, “Analytical
Formulation for the Resonant Frequency of Split Rings,” Journal of Applied
Physics 105, 014903 ,2009.
15. B. Sauviac, C. R. Simovski, S. A. Tretyakov, “Double Split-Ring Resonators:
Analytical Modeling and Numerical Simulations,” Electromagnetics, 24:317–
338, 2004, Taylor & Francis Inc.ISSN: 0272-6343 print/1532-527X
online,DOI: 10.1080/02726340490457890.
94
16. H. Chen, L. Ran, J. Huangfu, T. M. Grzegorczyk and J.Kong, “Equivalent
Circuit Model for Left-Handed Metamaterials,” Journal of Applied Physics
100, 024915, 2006.
17. I. Gil, J. Bonache, M. Gil, J. García-García, and F. Martín, R. Marqués,
“Accurate Circuit Analysis of Resonant-Type Left Handed Transmission
Lines with Inter-Resonator Coupling,” Journal of Applied Physics 100,
074908, 2006.
18. T. J.Yen, W.J. Padilla , N. Fang, D.C. Vier , D.R.Smith, J. B. Pendry ,D.N.
Basov and X. Zhang , “Terahertz Magnetic Response from Artificial
Materials,” Science 303, 1494, 2004.
19. J. Zhou, T. Koschny and C. M. Soukoulis, “Magnetic and Electric Excitations
in Split Ring Resonators,” Optics Express 17881, Vol. 15, No. 26, 24
December 2007.
20. S.G. McMeekin, and A.Z Khokhar, L. Basudev, R.M. De La Rue, and N.P
Johnson, (2007) “Analysis of Resonant Responses of Split Ring Resonators
Using Conformal Mapping Techniques,” In, Kuzmiak, V.and Markos, P. and
Szoplik, T., Eds. Metamaterials II, 16-18 April 2007 Proceedings of SPIE--the
International Society for Optical Engineering Vol 6581, Prague.
21. D. M. Pozar, “Microwave Engineering,” Page 187, 2005.
22. M.
T.
Thompson,
“Inductance
Calculation
Techniques,
Part
II:
Approximations and Handbook,” 1999.
23. G. Turhan-Sayan, “Unpublished Notes on the Equivalent Circuit Modeling for
SRR Arrays,” METU, 2009.
95